Suitability of a Commercial Software Defined Radio System for ...

Suitability of a Commercial Software Defined Radio
System for Passive Coherent Location
Aadil Volkwin
A dissertation submitted to the Department of Electrical Engineering,
University of Cape Town,
in fulfilment of the requirements for the degree of
Master of Science in Engineering.
Cape Town, May 2008

Dedicated to:
My Parents,
my Sister and my Wife.
2

Declaration
I declare that this work done is my own, unaided work. This dissertation is being submitted to
the Department of Electrical Engineering, University of Cape Town, in fulfilment of the
requirements for the degree of Master of Science in Engineering. It has not been submitted for
any degree or examination in any other university.
................................................................................
Signature of Author
Cape Town
2008
3

Abstract
This dissertation provides a comprehensive discussion around bistatic radar with specific
reference to PCL, highlighting existing literature and work, examining the various performance
metrics. In particular the performance of commercial FM radio broadcasts as the radar waveform
is examined by implementation of the ambiguity function.
The FM signals show desirable characteristics in the context of our application, the average
range resolution obtained is 5.98km, with range and doppler peak sidelobe levels measured at
-25.98dB and -33.14dB respectively.
Furthermore, the SDR paradigm and technology is examined, with discussion around the design
considerations. The USRP, the TVRx daughterboard and GNURadio are examined further as a
potential receiver and development environment, in this light.
The system meets the low cost ambitions costing just over US$1000.00 for the USRP
motherboard and a single daughterboard. Furthermore it performs well, displaying desirable
characteristics, The receiver's frontend provides a bandwidth of 6MHz and a tunable range
between 50MHz and 800MHz, with a tuning step size as low as 31.25kHz. The noise
characterisation of the receiver reveals a NF of 10dB, a sensitivity of -105dB and a dynamic
range of 62dB.
Finally, the investigation into the stability of the daughterboard frontend oscillators due to ageing
effects is shown to be steady with acceptable levels of variation, showing a fractional frequency
variation from 2.3 to 0.8 parts per 10 million and a maximum frequency drift of 4Hz.
4

Acknowledgements
All praises and thanks to the almighty for all the blessings and abilities that have been afforded
to me.
Further gratitude is expressed to:
My supervisor, Prof. M.R. Inggs for his advice, support and encouragement.
Dr. Yoann Paichard, for invaluable technical advice in developing and interpreting various
aspects of the research.
To the RRSG, in particular Regine Lord for administrative support, Lance Williams, Jonathan
Ward and Gunther Lange for various hardware assistance.
To members of the GNURadio mailing list, in particular Matt Ettus and Firas Abbas.
Finally, to my family and friends for their encouragement, support and patience .
5

Contents
Declaration.......................................................................................................................................3
Abstract............................................................................................................................................4
Acknowledgements..........................................................................................................................5
Contents........................................................................................................................................... 6
List of Figures..................................................................................................................................8
List of Tables................................................................................................................................. 10
Nomenclature.................................................................................................................................11
Chapter 1........................................................................................................................................13
Introduction................................................................................................................................... 13
1.1 Background........................................................................................................................ 13
1.2 Plan of Development..........................................................................................................14
Chapter 2........................................................................................................................................20
Passive Coherent Location............................................................................................................ 20
2.1 System level description of PCL........................................................................................20
2.2 Advantages and Disadvantages of PCL Systems...............................................................25
2.3 History and Examples of PCL Systems............................................................................. 26
2.4 Applications of PCL.......................................................................................................... 28
2.5 Typical Illuminators...........................................................................................................28
2.6 Relevant Equations .......................................................................................................... 30
2.7 Summary ........................................................................................................................... 40
Chapter 3........................................................................................................................................41
USRP and GNURadio................................................................................................................... 41
3.1 Introduction ......................................................................................................................41
3.2 Software Defined Radio ....................................................................................................41
3.3 Anatomy of a SDR Receiver..............................................................................................43
3.4 Universal Software Radio Peripheral.................................................................................44
3.5 GNURadio......................................................................................................................... 52
3.6 Summary............................................................................................................................ 55
Chapter 4........................................................................................................................................57
Receiver Characterisation..............................................................................................................57
4.1 Introduction ......................................................................................................................57
4.2 Test System Overview ...................................................................................................... 57
4.3 Receiver Bandwidth...........................................................................................................58
4.4 Gain Range and Control.....................................................................................................59
4.5 Multiplexer Usage and Data Interleaving.......................................................................... 61
4.6 Frequency Range and Control........................................................................................... 62
4.7 Spurious Artifact ...............................................................................................................63
4.8 Receiver Noise Figure .......................................................................................................64
4.9 Sensitivity.......................................................................................................................... 66
4.10 Dynamic Range, Minimum Detectable Signal and Gain Compression...........................67
4.11 Summary.......................................................................................................................... 71
6

Chapter 5........................................................................................................................................73
Ambiguity Plot Analysis............................................................................................................... 73
5.1 Introduction ......................................................................................................................73
5.2 System Overview............................................................................................................... 73
5.3 Algorithm Concept.............................................................................................................75
5.4 Results................................................................................................................................76
5.5 Summary............................................................................................................................ 82
Chapter 6........................................................................................................................................84
Oscillator Stability.........................................................................................................................84
6.1 Introduction ......................................................................................................................84
6.2 Quartz Crystal Oscillators ................................................................................................85
6.3 Characterisation of stability .............................................................................................. 86
6.4 System Overview .............................................................................................................. 92
6.5 Results ..............................................................................................................................94
6.6 Summary............................................................................................................................ 95
Chapter 7........................................................................................................................................96
Conclusions and Recommendations..............................................................................................96
Bibliography ................................................................................................................................ 98
Appendix A ................................................................................................................................100
Ambiguity Analysis and Results................................................................................................. 100
Appendix B ................................................................................................................................116
Datasheet and Sentech Table.......................................................................................................116
7

List of Figures
Figure 1.1: PCL block diagram..................................................................................................... 15
Figure 1.2: Software Radio block diagram....................................................................................15
Figure 1.3: Experimental setup of the USRP for the characterisation tests...................................16
Figure 1.4: USRP with two TVRx boards, reference channel and target channel........................ 18
Figure 2.1: PCL system diagram................................................................................................... 22
Figure 2.2: Variation of RCS and angular width of a forward scatter target against frequency... 32
Figure 2.3: A target echo detected after the transmission of the second pulse results in ambiguous
ranges at B and C...........................................................................................................................36
Figure 3.1: USRP and daughterboard block diagram....................................................................45
Figure 3.2: USRP and daughterboards.......................................................................................... 46
Figure 3.3: AD9862 ADC block Diagram.....................................................................................47
Figure 3.4: USRP and FPGA block Diagram................................................................................48
Figure 3.5: DDC block Diagram................................................................................................... 49
Figure 3.6: USRP transmit chain block Diagram.......................................................................... 50
Figure 3.7: Signal flow from a single TVRx daughterboard to DDC0......................................... 51
Figure 3.8: GNURadio application structure.................................................................................53
Figure 4.1: Experimental setup used in characterisation tests.......................................................58
Figure 4.2: Block diagram showing TVRx gain stages.................................................................59
Figure 4.3: Variation of voltage applied to RF AGC against RF Gain......................................... 60
Figure 4.4: Variation of voltage applied to IF AGC against IF Gain............................................60
Figure 4.5: multiplexer setup for two TVRx to DDC0 and DDC1............................................... 61
Figure 4.6: Spurious signal generated by USRP on startup...........................................................63
Figure 4.7: closer look at the spurious signal generated by USRP on startup...............................64
Figure 4.8: Cascaded components................................................................................................. 65
Figure 4.9: Variation of receiver sensitivity..................................................................................67
Figure 4.10: MDS, 1 dB compression point and dynamic range...................................................68
Figure 4.11: Spurious Free Dynamic Range..................................................................................69
Figure 4.12: Input power against output power, showing receiver dynamic range.......................70
Figure 4.13: TVRx and ADC dynamic range comparison............................................................ 71
Figure 5.1: Experiment setup to capture signals for ambiguity function analysis.........................74
Figure 5.2: Block Diagram of experimental setup.........................................................................74
Figure 5.3: Signal processing algorithm........................................................................................75
Figure 5.4: Ambiguity plot of Rad5 transmission......................................................................... 77
Figure 5.4a: Zero doppler cut through ambiguity plot.................................................................. 77
Figure 5.4b: Zero delay cut through ambiguity plot......................................................................78
Figure 5.5a: Variation of signal range resolution with time..........................................................81
Figure 5.5b: Variation of signal range resolution with time..........................................................82
Figure 6.1: Determining the fractional frequency......................................................................... 88
Figure 6.2: Dependence of stability measurement on time window............................................. 89
8

Figure 6.3: measurement process.................................................................................................. 91
Figure 6.4: Block diagram of experimental setup......................................................................... 92
Figure 6.5: Experiment setup for stability tests.............................................................................93
Figure 6.6: Fractional frequency variation of the daughterboards................................................ 94
Figure 6.7: Allan Variance of the fractional frequencies.............................................................. 95
Figure A.1: Ambiguity plot of Radio 2000 transmission............................................................ 100
Figure A.1 a: Zero doppler cut through ambiguity plot...............................................................101
Figure A.1 b: Zero delay cut through ambiguity plot..................................................................101
Figure A.2: Ambiguity plot of Classic FM transmission............................................................ 103
Figure A.2a: Zero doppler cut through ambiguity plot................................................................103
Figure A.2b: Zero delay cut through ambiguity plot...................................................................104
Figure A.3: Ambiguity plot of UmhloboFM transmission..........................................................105
Figure A.3a: Zero doppler cut through ambiguity plot................................................................106
Figure A.3b: Zero delay cut through ambiguity plot...................................................................106
Figure A.4: Ambiguity plot for RGHP transmission...................................................................108
Figure A.4a: Zero doppler cut through ambiguity plot................................................................108
Figure A.4b: Zero delay cut through ambiguity plot...................................................................109
Figure A.5: Ambiguity plot of RSGR transmission.................................................................... 110
Figure A.5a: Zero doppler cut through ambiguity plot................................................................111
Figure A.5b: Zero delay cut through ambiguity plot...................................................................111
Figure A.6: Ambiguity plot of SAFM transmission....................................................................113
Figure A.6a: Zero doppler cut through ambiguity plot................................................................114
Figure A.6b: Zero delay cut through ambiguity plot...................................................................114
9

List of Tables
Table 5.1: Bandwidth Variation of RAD5 waveform................................................................... 78
Table 5.2: Summary of waveform ambiguity function performances...........................................83
Table A.1: Bandwidth variation of Radio 2000 waveform......................................................... 102
Table A.2: Bandwidth variation of Classic FM waveform..........................................................104
Table A.3: Bandwidth variation of Umhlobo FM waveform......................................................107
Table A.4: Bandwidth variation of RGHP waveform................................................................. 109
Table A.5: Bandwidth variation of RSFR waveform.................................................................. 112
Table A.6: Bandwidth variation of SAFM waveform.................................................................115
10

Nomenclature
AGC
ADBF
CFAR
CIC
CW
DBS
DDC
DOA
DPI
DSP
FPGA
FSF
GNU
GPIF
GPS
GSM
IP3
LO
MDS
MTI
NCO
PCL
PGA
PPM
RCS
SAR
SDR
SFDR
Automatic Gain Control
Adaptive Beamformer
Constant False Alarm Rate
Cascaded Integrator Comb Filter
Continuous Wave
Direct Broadcast by Satellite
Digital Down Converter
Direction Of Arrival
Direct Path Interference
Digital Signal Processors
Field Programmable Gate Arrays
Free Software Foundation
GNU Not Unix
General Purpose Interface
Global Positioning System
Global System for Mobile
Third Order Intercept Point
Local Oscillator
Minimum Detectable Signal
Moving Target Indicator
Numerically Controlled Oscillator
Passive Coherent Location radar
Programmable Gain Amplifier
Part Per Million
Radar Cross Section
Synthetic Aperture Radar
Software Defined Radio
Spurious Free Dynamic Range
11

SINAD
SIR
SNR
USRP
Signal to Noise and Distortion ratio
Signal to Interference Ratio
Signal to Noise Ratio
Universal Software Radio Peripheral
12

Chapter 1
Introduction
The University of Cape Town (UCT) Radar Remote Sensing Group (RRSG) is currently
engaged in the research and development of an air traffic control system that utilises passive
coherent location (PCL) radar.
Thus the purpose of this dissertation is to investigate the performance of the Universal Software
Radio Peripheral (USRP) and the GNU Not Unix (GNU) Radio software radio development
environment as a low cost PCL receiver and additionally to investigate and comment on the
performance of analogue FM radio broadcast signals as the radar waveform.
1.1 Background
The first radar systems built were bistatic in nature, with transmitter and receiver in separate
locations. An example of this is the Klein Heidelberg, developed by the Germans during the
second world war, that used the British Chain Home radars as illuminators. Technological
developments in the form of the duplexer in 1936 allowed the transmitter and receiver to share a
common antenna and therefore the same site [34].
This provided the means for simplicity of operation as well as savings on cost and space and
hence monostatic radar dominated radar research and design.
Bistatic radar does have advantages over monostatic radar, however, their disadvantages; in
particular the complexity of transmitter/receiver synchronisation and antenna beam pointing had
previously outweighed their advantages and hampered the advancement of bistatic radar.
Technological improvements in the shape of high speed digital signal processors (DSP), phased
array antennas and the deployment of global positioning system (GPS) satellite navigation
systems, which can be used for synchronisation, have provided the means to mitigate the
complexities of bistatic radar and thus fuelling the current wave of interest in bistatic radar
systems.
Interest in bistatic radar varies on a period of fifteen years, currently we are at a peak of that
cycle; with particular interest in PCL techniques [35], using 'illuminators of opportunity' such as
existing radar transmissions and broadcast and communication signals.
13

Of the illuminators of opportunity available in the environment, broadcast transmitters present
themselves as the most attractive for surveillance purposes, owing to their inherent properties of
high powers of transmission and widespread coverage.
If such signals are to be used on the basis of PCL, it is necessary to know their behaviour in
terms of their ambiguity function [36] i.e. resolution and sidelobe levels in range and doppler,
and the effect of range and doppler ambiguities.
Over the last ten years the enhancement of semiconductor technologies in performance
capability and cost has led to the emergence of Software Defined Radio (SDR) as a mainstream
technology. SDR is defined as a collection of hardware and software technologies that enable
reconfigurable system architectures for wireless networks and user terminals. SDR provides an
efficient and comparatively inexpensive solution to the problem of building multi-mode, multiband,
multi-functional wireless devices that can be adapted, upgraded or enhanced by using
software upgrades [31].
The obvious flexibility and costs benefits offered by SDR systems provide the impetus for the
choice to explore the USRP and GNURadio as a radar receiver in a bistatic PCL context.
1.2 Plan of Development
Chapter 1:
This chapter, which provides a brief background to the subject of the dissertation and motivation
for the research pursued.
Chapter 2:
This chapter reviews the published material on PCL radar techniques to provide the background
information to the techniques used within the research. This chapter begins by defining PCL
radar and its uses. The discussion then evolves describing the various advantages and
disadvantages for the system. Thereafter, the history of bistatic radars is discussed, ranging from
the first use of these systems to modern day systems. The discussion progresses further to
highlight previous work in the development of a PCL radar system and examines the
characteristics that motivate the choice of commercial FM radio broadcasts as a potential radar
waveform. Various applications of PCL radars were described, both for military uses as well as
non military uses. These applications were presented to show how bistatic radars have been of
use to society. Some of the enhanced techniques of bistatic and PCL systems are also discussed
with regards to the performance of these systems. The literature reviewed provides the necessary
background understanding and sets up the context in which this research finds purpose.
The following image depicts the geometrical setup of a PCL system.
14

Figure 1.1: PCL block diagram
Chapter 3:
Software radio is a system which turns radio hardware problems into software problems. The
fundamental characteristic of software radio is that software defines the transmitted waveforms
and software demodulates the received waveforms. Thus a software radio can tune to any
frequency band and receive any modulation across a large frequency spectrum. A block diagram
of the basic SDR structure is shown below.
Figure 1.2: Software Radio block diagram
15

Additionally, performing a significant amount of the signal processing on reprogrammable
hardware permits the radios to receive and transmit a new form of radio protocol simply by
running an alternative configuration of the hardware. This is unlike most radios in which the
processing is done with either analogue circuitry alone or combined with digital chips.
The chapter explores the USRP and GNURadio as development tools of this technology, in the
light of this. Through this it is found that the USRP provides a flexible and powerful FPGA, the
Altera cyclone, as the processing unit, which implements a great deal of the signal processing in
the receive and transmit chains. Furthermore, the USRP hosts two Analog Devices AD9862
chips, each providing a two ADCs sampling at a rate of 64 Msamples/second with 12 bits of
resolution and two DACs operating at 128 Msamples/second with 14 bits of resolution.
Moreover, the AD9862 chips provide two auxiliary ADCs and three auxiliary DACs which can
provide further flexibility in computation and control of external daughterboard components.
These daughterboards provide front end flexibility allowing for the down-conversion and thus
manipulation of the radio spectrum.
GNURadio is seen to be a powerful toolkit that implements numerous complex signal processing
elements and additionally interfaces with the USRP. Thus, GNURadio and the USRP combine to
provide a powerful and widely versatile toolset for the development of a variety of radio devices.
Chapter 4:
Often the most critical component of a wireless system is its receiver, the purpose of which is to
extract, reliably, the desired signal from the various sources of signals, interference and noise.
This chapter presents the results of experiments that examine and characterise the performance
of the USRP, TVRx daughterboard, which is a complete VHF and UHF receiver system based
on a TV tuner module and GNURadio as a PCL receiver.
Figure 1.3: Experimental setup of the USRP for the characterisation tests
A number of the TVRx and USRP receiver system parameters as well as the GNURadio
interface to the controllable aspects are examined. The receiver's frontend provides a bandwidth
of 6MHz, which far exceeds the performance requirements of our application. and a tunable
range between 50MHz and 800MHz, with a tuning step size as low as 31.25kHz. This is
sufficient for the manipulation of FM Broadcast signals as the radar waveform and provides
further functionality and versatility to incorporate additional areas of the frequency spectrum if
desired, i.e. the television video and sound carriers.
16

The multiplexer implemented in the FPGA and the ability to directly control its operation
through GNURadio further extends the receiver's flexibility in its digital manipulation of signals
through it. The noise characterisation of the receiver reveals a NF of 10dB, a sensitivity of
-105dB and a dynamic range of 62dB. Furthermore, it is observed that when the GNURadio
application is started, when the USRP is setup as a source and includes the occasion when the
application is stopped and restarted, a spurious spike occurs at the beginning of the data set. This
phenomenon is unexplained but noted by the USRP developers. The effects of this are mitigated
by simply cutting out that part of the data set.
Thus the USRP with the TVRx and GNURadio show performance characteristics suitable for
our application.
Chapter 5:
This chapter presents the measurements of the ambiguity functions of off-air signals that will be
considered for the PCL system and comments on their form and usefulness by demonstrating the
waveform variability and its effect on range and doppler resolution in addition to detection
performance.
From the results it is evident that the measured FM Broadcast signals due to the randomness in
their modulation, exhibit noise like characteristics and behaviour. This is observed as the
ambiguity function plots can be approximated by the ideal thumbtack ambiguity function,
therefore providing the radar with unambiguous range and doppler information. However, it is
additionally clear that the ambiguity function depends largely on the modulating format.
Thus there is a variation in performance between the pop music and dance music channels and
the rock and classical music channels which exhibit a slightly degraded performance. Speech
content shows the poorest performance though, as the performance does vary significantly with
time, especially during pauses in words due to the low spectral content. This variation in
performance does have a significant impact on the radar's potential detection ability and will be
one that does vary with time, as seen from Equation 2.6 and Table A.6, this variation translates
to a change in processing gain of 15dB.
Table 5.2 summarises the ambiguity function performance of the measured signals.
Signal
Table 5.2: Summary of waveform ambiguity function performances
Range Resolution
(km)
Effective
Bandwidth
(kHz)
Peak range
Sidelobe level
(dB)
Peak doppler
Sidelobe level
(dB)
RAD5 3.38 44.3 - 30 - 27
Radio 2000 7.94 18.9 - 18 - 34
Classic FM 7.98 18.8 - 30 - 33
Umhlobe 3.33 45.0 - 31 - 25
RGHP 2.36 63.6 - 30 - 37
RSGR 9.32 16.1 - 18 - 36
SAFM 7.54 19.9 - 20 - 40
Average 5.98 32.37 -25.29 -33.14
17

Chapter 6:
A limiting factor in the performance of radar applications is the ability of the radar's frequency
reference to maintain timing accuracy. Chapter 6 thus is a presentation of experimental
measurements aimed at determining the stability of the local oscillators present in the TVRx
frontend daughterboards of the USRP and hence the PCL receiver.
Figure 1.4: USRP with two TVRx boards, reference channel and target channel
The local oscillator in the TVRx daughterboard of the USRP is a 4MHz quartz crystal oscillator.
It is for this reason that the properties of crystal oscillators and the associated performance
parameters of interest were investigated. These parameters include accuracy, reproducibility and
stability. It was found that crystal oscillators provide good performance at a reasonable price and
dominate the field of frequency sources [19].
Furthermore, the techniques associated with the specification of stability were presented,
including measurement of the frequency deviation and the Allan variance. These tests concluded
that the crystal ageing effect is reduced over the measurement period of thirty minutes, showing
a variation from 2.3 to 0.8 parts per 10 million. In order to qualify the results of this chapter a
5% error in the calculated doppler shift is set as the maximum acceptable limit of frequency
shift. Thus a frequency drift of no greater than 5Hz is set as the requirement of the system.
Chapter 7:
This chapter presents the conclusions and recommendations of future work.
In particular the ambiguity function is examined and the performance of commercial FM radio
broadcasts as the radar waveform is determined and summarised by Table 5.2, reproduced here.
18

Signal
Table 5.2: Summary of waveform ambiguity function
Range Resolution
(km)
Effective
Bandwidth
(kHz)
Peak range
Sidelobe level
(dB)
Peak doppler
Sidelobe level
(dB)
RAD5 3.38 44.3 - 30 - 27
Radio 2000 7.94 18.9 - 18 - 34
Classic FM 7.98 18.8 - 30 - 33
Umhlobe 3.33 45.0 - 31 - 25
RGHP 2.36 63.6 - 30 - 37
RSGR 9.32 16.1 - 18 - 36
SAFM 7.54 19.9 - 20 - 40
Average 5.98 32.37 -25.29 -33.14
The analysis of the FM broadcast signal ambiguity plots reveal the attractive performance of the
signals as they in general tend toward the idealised thumbtack. However, this is greatly
dependant upon the instantaneous modulating content, revealing highly degraded performance in
the case of a signal with low spectral content such as speech with many pauses and breaks.
Furthermore, the SDR paradigm and technology is examined, with discussion around the design
considerations. The USRP, the TVRx daughterboard and GNURadio are examined further as a
potential receiver and development environment, in this light.
GNURadio, is found to be a powerful toolkit that implements numerous complex signal
processing elements and additionally interfaces with the USRP that provides flexible and
powerful computing capabilities. Thus combining to provide a powerful and widely versatile
toolset for the development of a variety of radio devices.
The system meets the low cost ambitions costing just over US$500.00 for the USRP
motherboard and a single daughterboard. The receiver's frontend provides a static bandwidth of
6MHz and a tunable range between 50MHz and 800MHz. The noise characterisation of the
receiver reveals a NF of 10dB, a sensitivity of -105dB and a dynamic range of 62dB.
Finally, the investigation into the stability of the daughterboard frontend oscillators due to ageing
effects is shown to be steady through examination of the fractional frequency variation as well as
the Allan variance, showing a fractional frequency variation from 2.3 to 0.8 parts per 10 million
and a frequency drift no greater than 4Hz. This falls within the maximum allowance of 5Hz,
calculated in section 6.1.
Appendix A:
Ambiguity Analysis Results.
Appendix B:
Relevant datasheets and Sentech transmission table.
Bibliography
19

Chapter 2
Passive Coherent Location
Conventional (dedicated) radar systems are comprised of a dedicated transmitter and receiver.
In a passive radar system, however, there is no dedicated transmitter. Instead the receiver uses
third-party non cooperative transmitters in the environment. PCL systems are bistatic in nature
and thus measure the difference in time of arrival between the signal directly from the
transmitter and the signal reflected off the target. This allows the bistatic range of the object to
be determined. In addition to the bistatic range, in order for the the location, direction and
velocity of the target to be calculated the bistatic doppler shift of the reflected signal and its
direction of arrival is determined.
Clearly, in a dedicated radar system, the the transmitter and its location are directly under the
control of the radar engineer, hence timing information related to the transmission of the pulse
and the form of the transmitted waveform are exactly known. This allows the target range to be
easily calculated and for a matched filter to be used to achieve an optimal signal to noise ratio
(SNR) in the receiver.
A passive radar however, does not have control over the transmitter, its location, the nature of
the transmitted waveform or timing information related to the transmission and therefore must
use a dedicated receiver “reference channel” to monitor each transmitter being exploited.
2.1 System level description of PCL
The purpose of this section is to provide the reader with insight into the workings and
implementation of the overall system, without providing an overwhelming detail of theory. This
will enable the reader to understand the context within which the later sections fit.
20

A passive radar would typically consist of the following processing steps:[33]. This is illustrated
in Figure 2.1
●
●
●
●
●
●
●
●
Reception of the direct signal from the transmitter(s) and from the surveillance region on
dedicated low-noise, linear, digital receivers.
Digital beam forming to determine the DOA of signals and spatial rejection of strong inband
interference.
Adaptive filtering to cancel any unwanted direct signal returns in the surveillance
channel(s).
Transmitter-specific signal conditioning.
Cross-correlation of Doppler-shifted copies of the reference channel with the surveillance
channel(s) to determine target bistatic range and Doppler.
Detection using a constant false alarm rate (CFAR) scheme.
Association and tracking of object returns in range/doppler space, known as “trajectory
tracking” typically employing a Kalman filter.
Association and fusion of line tracks from each transmitter to form the final estimate of
an object’s location, heading and speed.
21

Figure 2.1: PCL system diagram [33]
2.1.1 Receiver system
It is essential that the PCL receiver should have a low noise figure, high dynamic range and high
linearity. This is as a result of the nature of PCL systems using illuminators of opportunity with
continuous wave signals, wherein they must detect very small target returns in the presence of
continuous and significant interference.
In order to achieve the processing gain necessary to detect these weak target returns in a
background of noise and interference it is necessary to achieve an equivalent to the optimal
matched filter processing used in conventional radar systems.
22

Passing target echoes through a matched filter is equivalent to the correlation of the radar echo
with a delayed replica of the transmitted signal which is obtained via the reference channel
mentioned earlier.
The greatest source of interference and hence limitation on the system performance is the Direct
Path Interference (DPI) received from the transmitter being used to detect aircraft. This
unwanted DPI signal correlates perfectly with the reference signal and produces range and
doppler sidelobes that are several orders of magnitude greater than the echoes that are sought
[18].
To detect anything but the closest of targets, it is necessary to remove this signal,by both angular
nulling with the antenna and adaptive echo cancellation in the receiver. These strategies are
further elucidated in the sections to follow.
However, eventually the dynamic range of the receiver limits the cancellation and so the
principal limitation on system performance lies with the analogue to digital converter (ADC)
technology [16].
Further interference is due co-channel interference coming from other transmissions operating at
the same frequency within a single frequency network. This has a similar effect to the DPI.
In addition, the signal is always received against a background of reflections from the surface,
i.e. clutter, and could be buried under these interferences. Their power density at the radar
antenna, in addition to correlation between the desired signal and the interferences, further
influences the system’s performance.
These effects highlight the importance of conducting a characterisation of the USRP receiver.
A more thorough treatment of these interferences and the methods used to mitigate them are
discussed in [18] [12] [13].
2.1.2 Digital beamforming
Passive radar systems use antenna arrays with several antenna elements and element-level
digitisation. This allows the direction of arrival of echoes to be calculated using standard radar
beamforming techniques, such as amplitude monopulse using a series of fixed, overlapping
beams or more sophisticated adaptive beamforming. Alternatively, some research systems have
used only a pair of antenna elements and the phase-difference of arrival to calculate the direction
of arrival of the echoes (known as phase interferometry) [15] [40].
2.1.3 Signal conditioning
Adaptive nulling is spatial filtering through an adaptive beamformer (ADBF), to null the
interference. To null the interference, the output power of the beamformer along the direction of
the interference must be minimized, whereas the output power along the desired signal’s
direction must be maximized. This may include high quality analogue bandpass filtering of the
signal, channel equalization to improve the quality of the reference signal, removal of unwanted
structures in digital signals to improve the radar ambiguity function or even complete
reconstruction of the reference signal from the received digital signal [18] [33].
23

2.1.4 Adaptive filtering
The principal limitation in detection range for most passive radar systems is the signal to
interference ratio (SIR), due to the large and constant direct signal received from the transmitter.
To remove this, an adaptive filter can be used to remove the direct signal in a process similar to
active noise control. This step is essential to ensure that the range/doppler sidelobes of the direct
signal do not mask the smaller echoes in the subsequent cross-correlation stage.
In a few specific cases, the direct interference is not a limiting factor, due to the transmitter being
beyond the horizon or obscured by terrain (such as with the Manastash Ridge Radar), but this is
the exception rather than the rule, as the transmitter must normally be within line-of-sight of the
receiver to ensure good low-level coverage [33].
2.1.5 Cross-correlation processing
The key processing step in a passive radar is cross-correlation. This step acts as the matched
filter to provide the necessary processing gain thereby maximising the SNR. The response to this
matched filter with respect to time and doppler shift of the carrier frequency from a normalized
version of the transmitted waveform is known as the ambiguity function.
This topic is treated with greater detail in section 2.6 of this chapter.
Most analogue and digital broadcast signals are noise-like in nature, and as a consequence they
tend to only correlate with themselves. This presents a problem with moving targets, as the
doppler shift imposed on the echo means that it will not correlate with the direct signal from the
transmitter. As a result, the cross-correlation processing must implement a bank of matched
filters, each matched to a different target doppler shift.
2.1.6 Target detection
Targets are detected on the cross-correlation surface by applying an adaptive threshold, and
declaring all returns above this surface to be targets. A standard cell-averaging CFAR algorithm
is typically used and determines the range and doppler of each target.
2.1.7 Trajectory tracking
At this stage in the processing the system has determined the range and doppler of a number of
targets. In order to further process the data it is necessary to associate this plot data with
individual targets and this is performed using a conventional Kalman filter. Most false alarms are
rejected during this stage of the processing.
2.1.8 Track association and state estimation
After having associated plots-to-targets, the range and doppler data for each target are processed
by a non-linear estimator to determine the target’s location, speed and heading. Use of a nonlinear
estimator allows optimum use of the doppler information in this tracking process.
24

2.1.9 Narrow band and CW illumination sources
The above description assumes that the waveform of the transmitter being exploited possesses a
usable radar ambiguity function and hence cross-correlation yields a useful result. Some
broadcast signals, such as analogue television, contain a structure in the time domain that yields
a highly ambiguous or inaccurate result when cross-correlated. In this case, the processing
described above is ineffective.
If the signal contains a continuous wave (CW) component, however, such as a strong carrier
tone, then it is possible to detect and track targets in an alternative way. Over time, moving
targets will impose a changing doppler shift and direction of arrival on the CW tone that is
characteristic of the location, speed and heading of the target. It is therefore possible to use a
non-linear estimator to estimate the state the of the target from the time history of the doppler
and bearing measurements.
The vision carrier of analogue TV signals has been shown to have successfully achieved this
[15]. However, target track initiation is slow and difficult, and so the use of narrow band signals
i.e. FM radio broadcasts is probably best considered as an ancillary to the use of illuminators
with better ambiguity surfaces. Later sections examine the merits of such a choice in greater
detail.
2.2 Advantages and Disadvantages of PCL Systems
PCL systems operate in bistatic and multistatic configurations, thus inherit all the associated
advantages and disadvantages.
Further additional advantages include:
●
●
●
●
Low procurement costs as they do not have any transmitter hardware.
Lower costs of operation and maintenance due to lack of transmitter and moving parts.
Covert operation, including no need for frequency allocations.
Physically small and hence easily deployed in places where conventional radars cannot
be.
● Rapid updates, typically once a second [33].
Disadvantages:
●
●
Lack of control over the form, nature and origin of the transmitters and signal waveform.
Performance limitations and complexity of deployment due to direct path interference,
co-channel interference, etc.
25

2.3 History and Examples of PCL Systems
Much of the early work published on PCL systems was conducted at University College London
(UCL) and was undertaken in the late 1970s.
A number of experimental bistatic systems have been built and evaluated by Schoenenberger
who designed and built a system using a UHF Air Traffic Control radar at Heathrow airport as
an illuminator, and investigated particularly the problems of synchronization between receiver
and transmitter [28]. A real-time co-ordinate correction scheme was also developed for this
system. Further developments included a digital beamforming array for pulse chasing
experiments and a coherent MTI system using clutter from stable local echoes as a phase
reference [4].
Subsequent work at UCL attempted to use UHF television transmissions as illuminators of
opportunity, to detect aircraft targets landing and taking off from Heathrow airport.
This work made use of the pulsed-like nature of parts of the television waveform for bistatic use
and showed the ability to receive clutter from the surrounding buildings. By implementing a two
pulse MTI canceller, off line, they were able to “track” moving targets. This system was
impractical in the sense that they required a special transmission waveform and only has a range
resolution of 1800m and range ambiguity of 9600m.
In attempt to improve the system performance, the use of a multi burst test pattern was
investigated. This once again yielded positive results for static objects, yet failed to resolve
moving targets.
This work illustrated that terrestrial television, in the time domain, was not suitable for use in a
bistatic configuration and that the use of a pulse radar transmitter is the best case, however when
the transmitter is not radar like, then the autocorrelation function of the transmitted waveform is
of paramount importance [15].
In addition to this, the transmit power should be proportionate to the coverage required. In the
case of complex modulation functions such as television, the calculation is made on the basis of
the power of that part of the spectrum used by the radar. Moreover, the radiation pattern of the
transmitter should be either omnidirectional or pencil-beam. Finally, the modulation bandwidth
must be commensurable with the required range resolution [12] .
Further work on television in the time domain using sophisticated correlation techniques which
were applied to Direct Broadcast by Satellite (DBS) TV signal was still not able to detect
moving targets at useful range. Thus suggesting that the problems associated with useful target
detection using DBS TV prevent it from being a practical approach to detecting airborne
targets[8].
Howland [15] developed a UHF forward scatter system based on television transmissions. A
forward scatter system is not able to provide range information, due to the nature of the bistatic
geometry, Thus a different approach was adopted, measuring direction of arrival (DOA) by
phase interferometry and doppler shift of the vision carrier of the television signal. The angle of
arrival of a target echo is related to the phase difference of the received signal at the surveillance
antennas by:
= 2 d
(2.1)
sin
26

where:
d
is the angle of arrival of a target echo
is the phase difference of the received signal
is the wavelength of the signal
is the distance between the dipoles
This work focused on developing signal processing techniques used for target tracking by means
of an extended Kalman filter algorithm and assumes that there is only one receiver system, and
one remotely located television transmitter. On its own the information extracted from a single
measurement of doppler shift and DOA of the target echo, is not useful. Therefore, a series of
measurements for doppler and DOA with respect to time were taken.
This choice was due to the unique association of the target echo's change in doppler shift and
DOA with its course and velocity. Howland's research bore similarity to the work done at UCL,
with the exception that his work was done in the frequency domain and exploited DOA and was
able to demonstrate tracking of aircraft targets at ranges in excess of 100km from the receiver.
Further work by Howland [16] investigated the use of FM radio broadcast transmissions, which
closely resembles his previous work based on television transmissions, employing doppler, range
and bearing data. Adaptive filtering was applied to two surveillance channels to reject the direct
transmitter signal interference. A conventional CFAR detection scheme was applied to range and
doppler information resulting from the matched filtering to determine the range and doppler of
each target. With only two surveillance channels the direction finding system again uses phase
interferometry to estimate the target bearing.
Target tracking was employed by implementation of a Kalman filter and nonlinear estimation to
determine the target’s location, speed and heading. Use of a nonlinear estimator allowed
optimum use of the doppler information in this tracking process. Via this approach Howland was
able to demonstrate tracking of aircraft at ranges in excess of 150km from the receiver.
The Manastash Ridge Radar is a system conceived and built by John Sahr of the University of
Washington, Seattle, for studies of the ionosphere [26]. It uses a single VHF radio transmitter as
illuminator, and a receiver separated from the transmitter by a large mountain range
(Mt.Rainier). The receiving system is based on a standard digitizer card and PC, and is
extremely simple and cheap, approximately US$15 000. Synchronization is achieved by GPS,
giving uncertainties of 100ns in timing which equates to 15 m in range and 0.01 Hz in doppler.
Although the purpose of the system is for ionospheric studies, it does detect aircraft targets at
ranges up to about 100km. Their system vividly demonstrates that high performance can be
achieved from simple and inexpensive PCL systems.
Silent Sentry developed by the Lockheed Martin company, based on multiple VHF FM radio and
television transmissions. It has demonstrated tracking of aircraft and space targets renewing
tracking targets can be completed eight times in a second. The database of the system has stored
in it, about 55,000 correlative FM broadcast and digital audio broadcast parameters and is
advertised as being applicable to: air surveillance and tracking in areas of limited coverage;
capable of tracking low flying, non-cooperative, slow moving targets; continuous total volume
surveillance of air breathing and ballistic objects; low acquisition and operations cost,
unattended remotely managed.
27

2.4 Applications of PCL
Reported applications of PCL include:[10]
●
air-space surveillance
● maritime surveillance
●
●
●
●
atmospheric studies and ionospheric studies
oceanography
mapping lightning channels in thunderstorms
monitoring radioactive pollution
There are also reports of algorithm development for interferometry, target tracking and target
classification [10]. This panoramic diversity of systems and applications is indicative of the
increasing importance of PCL systems.
2.5 Typical Illuminators
A wide variety of RF emissions in the form of TV and radio broadcasts in addition to terrestrial
and space based communications has resulted in a wide range of signal types available for
exploitation by passive radar. Furthermore, many such transmissions are at VHF and UHF
frequencies, which allows these parts of the spectrum not normally available for radar use, and at
which stealth treatment of targets may be less effective, to be used.
Typical broadcast service sources include:
●
●
●
●
●
●
Analogue TV signals
FM broadcast radio signals
Global System for Mobile (GSM) base stations
Digital audio broadcasts
Digital video broadcasts
Terrestrial HDTV
So Why FM radio
Of all the transmitters of opportunity available in the environment, the typical broadcast service
sources represent some of the most attractive for surveillance purposes, owing to their inherent
attractive properties of very broad coverage and relatively high transmitter powers.
28

Satellite signals have generally been found to be inadequate for passive radar use [33]. This is as
a result of their transmission powers being too low, or because the orbits of the satellites are such
that illumination is too infrequent. However, exceptions to this include the exploitation of
satellite based radar and satellite radio systems.
Digital audio transmitters emit signals at a much higher frequency than FM broadcast
transmissions, however FM broadcast transmissions are comparatively higher in their power.
Thus, the maximum detection ranges offered by FM broadcast transmissions are greater than that
of Digital audio [10]. Furthermore, the coverage of Digital audio is generally poor, although the
transmitter networks are expanding elsewhere, they are currently non-existent in South Africa.
Cellular phone base stations transmit at a rather low power and would seem therefore to be
limited in their performance and application. However, there is an extensive network and targets
could be tracked through such a network and hence the coverage may be extended greatly. This
however comes at the cost of increased system complexity. Furthermore, base stations
deliberately concentrate their emissions towards the ground and may not necessarily have good
coverage of higher altitude aircraft.
Analogue television transmitters, with high radiated powers and a pulse like waveform structure
present themselves as an obvious choice of illuminator. However, in spite of these instant points
of appeal, it has been shown that the waveform is far from suited for radar usage when used in a
conventional radar matched filtering approach [12]. In an alternative approach however, making
use of doppler and bearing information in echoes of the television video carrier it is possible to
track aircraft at ranges of up to 260km from the receiver and 150km from the transmitter [15].
This approach is referred to as narrow band processing as only a few kilohertz and thus small
percentage of the 5.5MHz television waveform bandwidth is processed to determine the target
doppler shifts. The disadvantages of this stem from the relatively low information content in the
doppler measurements and the system must observe the target’s doppler history for an extended
time before there is sufficient information to locate the target.
Moreover, the use of non-linear estimation techniques to calculate the target’s trajectory mean
that a good estimate of the target’s location must be initially available. In the case of a forward
scatter radar the unique geometry can be manipulated in order to determine an expression of the
target’s location [2].
In the general bistatic problem it is necessary employ global optimisation schemes [15]. The
former approach is limited in its operational applications and the latter is neither robust nor
computationally efficient.
FM broadcast transmissions have been used to probe the ionosphere [26] and therefore might be
expected to have useful height surveillance. Their high powers and good coverage make FM
broadcast radio transmissions particularly well suited to air target detection for both civil and
military applications. Moreover, they could be used for marine navigation in coastal waters
although clutter will be a more significant problem.
In PCL systems wideband processing is defined as the use of a receiver that has a bandwidth that
is comparable to that of the waveform being processed. A typical FM radio broadcast occupies a
bandwidth of approximately 100kHz, thus offering a potential range resolution of up to 1500m.
It is clear that the signal offers useful target ranging information. However, as a result of this
lower bandwidth available, range and bearing are a factor of ten or so worse than a conventional
microwave radar.
29

Doppler, on the other hand, is two or three orders of magnitude more accurate. This is due to the
extended integration times that are potentially achieved by passive radar. The doppler
information can be used to provide a resolution comparable to conventional radars and by
simultaneously using multiple transmitters the system can achieve target location accuracies that
may be even better.
However, the suitability of a signal for target location is governed by more than its bandwidth.
Of greater value and importance is the ability of the radar receiver to locate the target
unambiguously. For this we must compute its ambiguity function.
2.6 Relevant Equations
2.6.1 The Bistatic Radar equation
The starting point of an analysis of the performance of a system is the radar equation. The radar
equation for a bistatic system is derived in the same way as for monostatic systems.
P r
P n
= P t G t
4 r 1
2 ⋅ b ⋅ 1
4 r 2
2 ⋅G r 2
4 ⋅ 1
k T 0
B F ⋅L
(2.2)
where: [10]
P r
P n
P t
G t
G r
r 1
r 2
b
k
T 0
B
F
L ≤1
is the received signal power
is the receiver noise power
is the transmit power
is the transmit antenna gain
is the receive antenna gain
is the transmitter to target range
is the target to receiver range
is the bistatic radar cross section
is the signal wavelength
is Boltzmann's constant
is the noise reference temperature, 290 K (standard room temperature)
is the receiver effective bandwidth
is the receiver effective noise figure
are system losses
30

setting the values:
●
●
b
= m
r 1
= r 2
= r m
where:
m
is the monostatic radar cross section
r m
is the radar to target range
allows for the reduction of the bistatic range equation to the monostatic case.
In order to use the radar range equation appropriately and correctly to predict performance of a
radar system it is necessary to understand the correct value of each parameter to be used.
The transmit power P t of the various illuminators of opportunity exploited in PCL is high.
This is due to the fact that the intended receivers often have inefficient antennas and poor noise
figures, with the transmission paths far from line of sight. Thus the substantial transmit power is
necessary to overcome the inefficiencies. However, for example, the analogue TV broadcast has
pronounced ambiguities every 64 s and thus does not exhibit favourable ambiguity
properties and a more attractive ambiguity may be realised by using a portion of the signal
spectrum, at the expense of reduced power. Therefore, when employing the radar equation it is
necessary to consider only the portion of the signal spectrum that is used for the radar purposes.
This may not necessarily be the same as the power of the entire signal spectrum.
2.6.2 Bistatic Radar Cross Section
The likelihood of target detection and location is dependant on the spatially influenced bistatic
RCS and target dynamics, in addition to the radar design parameters. The use of conventional
processing techniques will result in the detection of targets in range, doppler and angle.
Relatively little has been published regarding the bistatic RCS of targets and this remains an area
for future research. However, early work, that resulted in the formulation of the bistatic
equivalence theorem, states that that the bistatic RCS is equal to the monostatic RCS at the
bisector of the bistatic angle β, reduced in frequency by the factor cos (β/2)[9].
This theorem however, is valid typically, for small angles, when the bistatic angle, β, is 5 O or
less. Therefore, above 5 O the target bistatic RCS will not in general be the same as the
monostatic RCS, although will be comparable for targets that have a low monostatic RCS and
due to shadowing that occurs in the monostatic geometry but not the bistatic geometry. In a
monostatic radar, phase interference from two or more targets causes a distortion of the echo
signal. This in turn results in the apparent phase centre of the radar reflection swinging between
the targets. This random movement of radar reflecting centres, leads to jittered angle tracking,
known as target glint [14]. The effect of target glint is reduced in the bistatic RCS region, thus
the bistatic RCS has the advantage of glint reduction.
As the bistatic angle in increased to 180 O the forward scatter region is encountered where the
target lies on the transmitter-receiver baseline. Within this region, the bistatic RCS can be many
times greater than that of the monostatic RCS. The magnitude of the target’s forward scatter
return does not depend on the material composition, therefore irregular shaped targets, and radar
absorbent materials used on targets are still able to be detected.
31

Any target illuminated by the transmitter on the baseline, when the targets dimensions are larger
than the transmitted wavelength, produces a shadow. This shadow region occurs on the opposite
side of the target from the transmitter. This maybe understood with reference to Babinet’s
Principle. Babinet’s principle is an approximation according to which the amplitude of near
forward scattering by an opaque, planar object is the same as that of an aperture of the same
shape and size in a perfectly conducting screen [10].
The forward scatter cross-section of a target with cross sectional area A is given by:
b
= 4 A2
2
(2.3)
Where the radiation wavelength, is assumed small compared to the target dimension.
The angular width of the scattered signal in the horizontal or vertical plane is given by:
b
= d
(2.4)
where:
d is the target linear dimension.
The dependence of b and b for a target of the size of a typical aircraft, is shown below.
Where A = 10m 2 and d = 20m.
Figure 2.2: Variation of RCS and angular width of a forward scatter target against frequency [10]
32

Angular width decreases with an increase in frequency. Thus the forward scatter illumination is
focused into an increasingly narrower beam. This implies that frequencies around VHF and UHF
are more likely to be optimum for exploiting forward scatter as target detection may be achieved
over an adequately wide angular range. Taking advantage of the forward scatter configuration,
however, comes at the expense of target doppler and position information. However, target
location can be estimated using a combination of doppler and bearing as explained in [15].
2.6.3 Bistatic Clutter
The received signal at any location is the vector RF sum of the direct signal and of the multipath
propagation [12]. Most of these multipath signal energies would result from reflection off fixed
objects, such as, buildings, other large objects, other aircraft, etc. This multipath propagation is
defined as clutter. Clutter, in general, is a cause of degradation in radar system performance, as
the target returns must compete with the clutter returns at the receiver. More often than not
clutter RCS are larger than target RCS [4].
Bistatic clutter is subject to greater variability than the monostatic case, because there are more
variables associated with the geometry [34], in addition to frequency and surface composition.
Relatively little work has been done in developing models for bistatic clutter and available
experimental data from the intended receiver location at UCT is minimal,and thus remains an
area worthy of future research.
It is important to know the clutter power returns at the receiver, in order to prevent the receiver
from saturating. Thus the work to be presented in chapter 4, characterising the receiver, will
provide the reader with an indication of the expected performance in this regard.
2.6.4 Integration Gain
Radar signals are considered to be coherent when the relative phases of the signals have a known
relationship, even though they may be separated in time. To improve the SNR a combination of
coherent and incoherent integration maybe implemented. Coherent integration is performed by
correlating the target echo with the reference signal and in essence, is the addition of a sequence
of M coherent radar signals where the signals are summed vectorially and is regarded as a single
waveform. Through this process the SNR will increase by a factor M, as given by Equation 2.5, a
derivation of this can be found in [24].
SNR m
= M × SNR 1
(2.5)
Where:
M
SNR m
is the number of coherent radar signals.
Is the the SNR of the average of M signals.
In the case of incoherent integration, these M pulses will not add in phase and will not improve
the SNR. Instead the signals are processed separately and the separate observations are then
combined, usually by means of averaging.
33

The relative roles of coherent and incoherent integration will be a function of the coherency of
the target echo and of competing noise and clutter signals.The maximum signal processing gain
achievable is given by:
G p
= B T max
(2.6)
Where:
B
T max
is the waveform bandwidth.
Is the maximum integration dwell time.
The effective waveform bandwidth of an FM radio transmission is known to vary with time,
clearly this will translate to a variation in signal processing gain.
A rule of thumb determining the maximum value for the integration dwell time or maximum
coherent processing time is given by:
T max =
{ A R
} 1/2
(2.7)
where:
A R
is the radial component of the target acceleration.
The integration dwell time is the length of time taken to make an observation with a radar set,
the integration dwell time is dependant on the waveform coherence and target dynamics, namely
target velocity and manoeuvrability.
T max
consequently, is additionally a time varying entity and care will need to be observed in
its choice.
2.6.5 Range Resolution and Maximum Unambiguous Range
The minimum distance necessary to separate two targets in order to distinguish between them
defines the range resolution.
For both monostatic and bistatic cases, adequate separation between the two target echoes at the
receiver is taken to be:
c⋅
(2.8)
2
where:
is the compressed radar pulse width.
34

Furthermore, (2.9)
= 1 B
and thus, Equation 2.8 can be restated as: c
(2.10)
2 B
If the time delay between the echoes from two targets is > the pulse width then the two
echoes are resolvable. However, if the targets are closer than their echoes merge and are
unresolvable.
In order to generate this separation at a bistatic receiver, two-point targets must lie on a bistatic
isorange contour with a separation R B . In the monostatic case, the distance between the two
circles are constant, unlike the bistatic case.
The equation for a monostatic system resolution is given by the equation:
R M
= c⋅
2
(2.11)
For a bistatic radar, separation between the ellipses depend on the bistatic angles. As the bistatic
angle increases, so does the distance between the isorange contours, and vice versa. Eventually
on an extended baseline, or long ranges, they become equispaced, and start to represent a
monostatic case. An approximation for R B is:[40]
R B
= R M
cos 2
(2.12)
= c⋅
2cos 2
The above equation is based on the monostatic case, reduced by a factor of cos 2 . This
relationship is common in additional bistatic properties. For example bistatic doppler is reduced
by the factor cos and when beta is small the bistatic RCS is equal to the monostatic RCS
2
reduced by the factor cos 2 .
In general any measurement made with a basic resolution of M, will have a root mean square
(RMS) error M given by:[24]
M
M ≃
(2.13)
2⋅SNR
Thus for the range resolution, the range error is given by:[24]
R ≃
c
2 B 2⋅SNR
(2.14)
Thus the range accuracy is limited by the pulse effective bandwidth and SNR.
35

It is clear from the above that the resolution and its accuracy is improved by using shorter pulses,
i.e. Large pulse bandwidths. However, there's a lower limit to pulse duration. Depending on
various parameters a minimum signal energy is required in order to detect a target echo.
The energy carried by the pulse is a product of transmitter power and pulse duration. Therefore,
cutting pulse duration in half, for instance, would necessitate doubling transmitter power in order
to keep the signal's energy constant. The problem with that is that transmitter power cannot be
increased at will because of cost and other constraints. Increasing the rate at which pulses are
transmitted by the radar will increase the mean power radiated and thus the energy radiated.
However, radars are designed so that a range counter starts at the transmission of the pulse, is
read out when an echo is detected, and is reset upon transmission of the next pulse. Thus when
pulses are transmitted so frequently that a pulse is transmitted before the previous pulse has
completed the round trip to the target and back. The resulting effect is an uncertainty in
determining the correct relationship between the transmitted and the echo pulses and the target
range becomes ambiguous. As depicted in Figure 2.3.
The range beyond which targets appear as second time around echoes is called the maximum
unambiguous range [29], which is given by:
R M
u
=
c
2⋅PRF
(2.15)
where:
PRF is the Pulse Repetition Frequency and is the rate at which pulses are transmitted
The corresponding bistatic unambiguous range is given by:
R T
R R
u
=
c
PRF
(2.16)
which lies on an isorange contour, of major axis length equal to equation (B.1). Further details
on unambiguous bistatic PRF can be found in [34] [24] [17].
Figure 2.3: A target echo detected after the transmission of the second pulse results in ambiguous
ranges at B and C.
36

2.6.6 Doppler Resolution
In order to accurately measure the speed of the target the doppler shift will be used. This is the
change in radio frequency of the signal, caused by the relative motions of the target and receiver.
With the target in motion, the radar to target range as well as the phase of the radar signal are
continuously changing.
This change in phase with respect to time is related to the doppler angular frequency and can be
further shown [29] that the doppler frequency shift is given by:
where:
f d
= 2v r
= 2v r f 0
c
(2.17)
f d
f 0
v r
is the doppler frequency shift
is the transmitted frequency
is the radial velocity of the target with respect to the radar
The definition of the bistatic doppler frequency shift is different to the monostatic case, above.
The change in radio frequency is a result of the rate of change of the total path length travelled
by the signal. In the monostatic case this change in path length is equal to a change in the targets
range. For a bistatic system the doppler shift is given by:
f d
= − 1
d R T
R R
dt
(2.18)
The expression indicates a decrease in doppler frequency for an increasing path length [24].
The doppler frequency shift provides the means to distinguish between stationary and moving
targets. However, is not a measure of the radial velocity of the target with respect to the radar as
with the monostatic case.
The velocity resolution, which is the ability to separate two targets separated in doppler is
synonymous with the doppler resolution and can be expressed by:
f d
= 1
PRI
(2.19)
The longer the duration of the signal, the finer the resolution will be and the more accurately the
doppler frequency shift can be determined.
From equation (2.19) above, the accuracy of the doppler resolution is given by:
f d
≃
1
PRI⋅ 2⋅SNR
(2.20)
37

Thus it is noted that the requirement for velocity resolution and accuracy implies long duration
waveforms whereas for range resolution and accuracy implies short waveforms. This paradox
must be considered during the signal processing and waveform design.
2.6.7 Ambiguity Function
The radar system's resolution, detection, measurement accuracy and ambiguity are defined by the
radar waveform [29]. The effect of the waveform on these parameters maybe determined upon
examination of the receiver output, which is optimum when designed on the matched filter
among others [29] [24] [22]. Should a target be present where it is expected to be, the receiver
output will peak to a maximum, in the absence of noise.
Thus the probability of detection is independent of the shape of the waveform and depends on
the ratio of the energy in the signal to the noise power per cycle of bandwidth. Range and
doppler accuracies are additionally dependant on this ratio, as previously discussed, additionally
are affected by the shape of the radar waveform as are ambiguity and resolution.
Ambiguity occurs when more than one choice for a particular parameter is available but only one
choice is expected. With respect to the matched filter receiver output, ambiguities arise when
peaks occur at values other than the expected value.
The effect of the transmitted waveform on the doppler and range can be produced upon
extension of the matched filter or autocorrelation function of the received signal. A three
dimensional plot of echo turn around time, doppler frequency and the output of the function
results in the ambiguity diagram which graphically presents the accuracy and ambiguity afforded
by the transmitted radar waveform. Additionally, to examine the waveform's potential to resolve
two targets of equal amplitude with different doppler and range [29].
The ambiguity plot is determined by the ambiguity function and as is given by:
∣ R R , f d ∣ 2 ∞
= ∣ ∫ s
−∞ t
t s ∗ 2
t
t−R R
exp[ j 2 f d
t ]dt ∣
(2.21)
where:
R R
f d
is the time delay (or range)
is the signal doppler shift
R R
, f d
is the ambiguity response at R R and f d
The ideal ambiguity diagram, also know as the thumbtack surface, would consist of a single peak
with a minuscule thickness at the origin and zero elsewhere [29]. This single spike prevents
ambiguities and permits the frequency and echo delay time to be determined simultaneously to a
high degree of accuracy. Furthermore, no matter their position, any two targets are resolvable.
However, the fundamental properties of the ambiguity function prevent this type of behaviour.
The main restrictions affecting the performance are that the maximum height of the function, and
therefore the peak at the origin are of a fixed height. In addition to this, the volume enclosed by
the function is fixed and finite.
38

Typically, the function will be spread about the origin and the maximum peak height will be no
greater than the fixed maximum amount thus the remaining energy is spread through the
remainder of the diagram which may result in one or more additional peaks, perhaps has high as
the one at the origin. The ambiguity diagram is centred at the origin, thus the response at
R R
, 0 is the response to reflections at a different range but at the same doppler as the
target. 0, f d
is the response to reflections at the same range as the target but at different
doppler shifts.
In PCL radar, there are two factors which affect the ambiguity function. These are the geometry
of the system and the instantaneous modulation of the reference signal.
The dependence of the ambiguity function of a bistatic radar on geometry has been evaluated by
Tsao et al [40] [9] [17] [11]. They show that in the bistatic arrangement, time delay is no longer
a linear function of target range and doppler is no longer a linear function of target velocity.
Thus, the form of the ambiguity function depends on the position and direction of motion of the
target.
Thus the bistatic ambiguity function of the signal
st is given by:
∣ R RH , R Ra , V H , V a , R , L ∣ 2 =
(2.22)
∣
∞
2
−∞
∫ s t
t− a
R Ra
, R
,Ls ∗ t
t− H
R RH
, R
, L
exp [ j 2 f dH
R RH
,V H
, R
, L−2 f da
R Ra
, V a
, R
, Lt ]dt∣
where:
R RH
R Ra
V H
V a
f dH
f da
R
L
is the hypothesised range from the receiver to the target
is the actual range from the receiver to the target
is the hypothesised radial velocity of the target with respect to the receiver
is the actual radial velocity of the target with respect to the receiver
is the hypothesised doppler frequency
is the actual doppler frequency
is the angle from the receiver to the target with respect to North
is the length of the baseline formed by the transmitter and receiver
The reference point of the geometry is assumed is assumed to be the receiver. The important
difference between this form of the ambiguity function and the form presented in Equation 2.21,
is that the geometrical layout and relationship between the transmitter, receiver and target have
been factored in. This can have a significant effect on the form of the ambiguity function and the
resulting range and doppler resolutions.
39

It is important to know the properties of these ambiguity functions with time, as the variation in
the form of the ambiguity function would determine the radar’s performance.
Thus investigation into the dependence of the ambiguity function on the instantaneous
modulation or programme content of the particular signal being broadcast forms, in part, the
objective of this dissertation.
2.7 Summary
This chapter introduces and explores the concept of PCL and notes that fundamentally, PCL
systems are bistatic in nature and thus measure the difference in time of arrival between the
signal directly from the transmitter and the signal reflected off the target. Furthermore, PCL
systems make use non-co-operative transmitters using 'illuminators of opportunity' such as
existing radar transmissions and broadcast and communication signals.
Some of the unique advantages and disadvantages that arise as a result of the geometry and the
use of illuminators of opportunity are listed. Although the performance of the bistatic
arrangement does not rival that of the monostatic geometry, there are some distinct applications
within the bistatic system which cannot be implemented by monostatic radars. An example
whereby bistatic systems are a major advantage, is using such a system to detect stealthy aircraft
and complete systems can be built on an extremely low budget.
The history and progression of development and research in PCL systems is explored and thus
highlighting the techniques developed by Howland, that express the viability with regards to the
tracking accuracy, as well as the distance at which targets were detected. Furthermore, work
presented by Baker and Griffiths examines the performance of a variety of potential waveforms
as the radar signal.
Of the illuminators of opportunity available in the environment, broadcast transmitters present
themselves as the most attractive for surveillance purposes, owing to their inherent properties of
high powers of transmission and widespread coverage.
If such signals are to be used on the basis of PCL, it is necessary to know their behaviour in
terms of their ambiguity function [36] i.e. resolution and sidelobe levels in range and doppler,
and the effect of range and doppler ambiguities. These signal processing techniques are outlined
at the end of the chapter, furthermore, the theory begins to highlight the purpose of work done in
measuring the the radar's frequency reference to maintain timing accuracy.
40

Chapter 3
USRP and GNURadio
3.1 Introduction
The continuum of innovation in semiconductor technology has led to the development of new
radio technologies. One such technology is software defined radio. Thus far, no standard
definition of software radio has been generated. This can be largely attributed to the nature of the
flexibility that software defined radios offer.
As the interest and proliferation of this technology pushes ever widening frontiers, the Free
Software Foundation (FSF) has developed and is maintaining a Software Defined Radio project
called GNURadio. This GNU project is a collection of software tools that can be used to
implement signal processing and radio applications on a PC using external USB based hardware,
known as the Universal Software Radio Peripheral (USRP) also developed as part of
GNURadio.
This chapter presents the research into the SDR paradigm and the associated design
considerations, thus providing the natural backdrop against which the performance of the USRP
and GNURadio are examined.
3.2 Software Defined Radio
However, SDRs do have characteristics that make them unique in comparison to other types of
radios. SDRs have the ability to be transformed through the use of software and redefinable
logic. This is achieved with general purpose digital signal processors (DSPs) or field
programmable gate arrays (FPGAs). Thus SDRs have the ability to go beyond simple single
channel, single mode transceiver technology with the ability to change modes arbitrarily because
the channel bandwidth, rate, and modulation are all flexibly determined through software. This is
unlike most radios in which the processing is done with either analogue circuitry or analogue
circuitry combined with digital chips.
41

As an example, a European GSM SDR device may load a codec to process American GSM
signals on detecting a change in the underlying service network, continuing to demodulate GSM
signalling with no change in the physical hardware [20].
In order to make full use of this concept, SDRs keep the signal in the digital domain for as much
of the signal chain as possible, digitizing and reconstructing close to the antenna, thereby
allowing digital techniques to perform functions done by analogue components as well as the
additional functions that are not possible in the analogue domain. To achieve this digitisation,
ADCs, in the receive chain and DACs, in the transmit chain, are employed. However, connecting
these components directly to the antenna introduces concerns regarding selectivity and
sensitivity.
Thus the need for a flexible analogue front end capable of translating a wide range of
frequencies and bands to that which the data converters are able to adequately process [25].
Some of the typical dynamic characteristics of an SDR include:
●
●
●
●
Channel Bandwidth
Data rates
Modulation type
Conversion Gain
This is exemplified in multiband radio systems, these systems are capable of operating on two or
more frequency bands sequentially or simultaneously. Typically multiple radios would be
required, each limited to operation in a specified band. In addition to this, multicarrier or
multichannel radios have the ability to simultaneously operate on more than one frequency at a
time, whether or not they fall in the same frequency band.
Multimode systems process several different kinds of standards and can be continuously
reprogrammed. Examples of such standards include AM, FM, GMSK, and CDMA. Furthermore,
a traditional radio determines the channel bandwidth with a fixed analogue filter. However, a
SDR determines the channel bandwidth using digital filters that can be altered. This provides
SDRs with the resources to vary the bandwidth dynamically, additionally, digital filters have the
potential to implement filters not possible in the analogue domain and are able compensate for
transmission path distortion [7]. These features are complex to implement using analogue filters.
3.2.1 Applications of Software Radio
Software Radio is finding use in an ever broadening spectrum of applications. Software radio
enable rapid prototyping and deployment of sophisticated wireless systems, making it ideal for a
plethora of research and development applications in academia, industry and defence.
To name just a few, software radio systems have been deployed in:
●
●
●
●
●
Public Safety and mine/underground communications
Battlefield, survivable and ad-hoc networks
Passive Coherent Radar
Cognitive Radio
Radio Astronomy
42

3.2.2 Software Radio development Tools and Frameworks
Several Software Radio environments are available. The common and favoured solutions
include:
●
●
●
●
●
C/C++
Matlab/Simulink/LabView
JTRS
OSSIE
GNURadio
This dissertation focuses its investigation on GNURadio.
3.3 Anatomy of a SDR Receiver
The first element in the receiver chain is the antenna. In a fully developed SDR the antenna is
additionally a programmable component. The antenna generally tends to be the weakest element
of the receiver. This is primarily so as the antenna has a bandwidth that is a small percentage of
the centre frequency and therefore, multiband operation is difficult. However, in instances where
single bands of operation are used, this is not a problem.
In the ideal scenario, the SDR would have its ADC connected directly to the antenna of the
receiver, this is in order to facilitate the implementation of as much of the system components in
the digital domain. This, however, is not a practical solution and some form of an analogue front
end must be used before the ADC in the receive path in order to conduct the appropriate
frequency translation.
The most common of these architectures is the super heterodyne architecture. In this instance the
midrange IF produced by the receiver allows the use of sharper cut off filters for improved
selectivity. Additionally higher IF gains are achieved through the use of IF amplifiers, improving
the sensitivity. The super heterodyne receiver uses two stages of conversion and is useful at
microwave frequencies in avoiding problems associated with Local Oscillator (LO) instability.
The super heterodyne receiver offers this performance over a range of frequencies, thus, by
combining wideband analogue techniques and multiple front ends facilitates the operation of the
receiver across different RF bands.
The direct conversion architecture, which converts the RF signal directly to an IF frequency of
zero, in less demanding applications, is seeing an increase in popularity. This architecture has the
advantage of being simpler to implement and less costly to build. This is so as they have no IF
amplifier, IF bandpass filter or IF LO required for the final down-conversion. Additionally, the
direct conversion approach does not generate any image frequencies as the mixer difference
frequency is effectively zero, while the sum frequency is easily filtered out.
However, this architecture has a major drawback in that the LO must have a very high degree of
precision and stability to avoid drifting of the received signal frequency. Currently, direct
conversion is found in user terminals for cellular communication [23]. Additionally, direct
conversion receivers are susceptible to various noise sources at DC, which create a DC offset
which may be larger than the signal itself and will degrade the SNR of the signal [1].
43

Band select filters are made use of in order to limit the range of input frequencies presented to
the high gain stage. This is in aid of minimising the effects of intermodulation distortion. In
addition, where intermodulation is not a problem, it is still possible that strong out of band
signals could limit the amount of potential gain in the following stages, resulting in limited
sensitivity. If all the signals input are of interest and filtering of the stronger signals is not
possible, it becomes necessary for the receiver to have a large dynamic range.
Mixers are used to translate the RF spectrum to a suitable IF frequency. Receivers may use a
number of mixer stages, each successively generating a lower frequency, as was described in the
super heterodyne architecture. In order to eliminate undesired images as well as other undesired
signals, filtering is made use of at each stage. The filtering should also be appropriate for the
application. For example a traditional single carrier receiver would generally apply channel
filtering through the mixer stages to help control the IP3 requirements of each stage [7].
A quadrature demodulator in addition to, or instead of, a mixer is employed in some receiver
architectures. This is in order to separate the I and Q components which undergo separate signal
conditioning. Due to the digital nature of the receiver this is not a problem. However, in the
analogue domain the signal paths must be perfectly matched, or I/Q imbalances will be
introduced, potentially limiting the suitability of the system.
An IF amplifier is often made use of in the form of an Automatic Gain Control (AGC). The goal
of the AGC is to use the maximum gain possible without saturating the the signal chain. If the
dynamic range in the receiver, which is a function of the ADC performance, is insufficient,
reduction in gain from a strong signal may cause weaker signals to be lost in the noise floor of
the receiver.
The ADC is used to convert the IF signal or signals into digital format for processing. The ADC
capabilities define the performance of the the SDR receiver. It is common place for the ADCs in
SDR receivers to be over specified as their applications are unknown prior to their selection and
the best available ADC is selected.
A number of options are available to implement the digital preprocessor. For very high sample
and data rates, this is usually implemented as a FPGA, which by nature are flexible in their
functions and range of parameters. An FPGA can, of course, be programmed for any function
desired. Typically, an FPGA would be programmed to perform the quadrature demodulation and
tuning, channel filtering, and data rate reduction among others. The FPGA is capable of
implementing and performing most signal processing tasks and control all of the features in the
other elements.
3.4 Universal Software Radio Peripheral
The USRP is a hardware platform that encompasses the software radio paradigm. It has been
designed to interface primarily with the GNURadio software, thereby providing a cheap, flexible
and high speed software radio prototyping test bed. The USRP is comprised of the motherboard,
which houses a FPGA for high speed signal processing, and interchangeable daughterboards
that cover different frequency ranges. The average cost of the USRP motherboard and a single
daughterboard is US$1000.00
44

The USRP provides several functions; digitization of the input signal, digital tuning within the IF
band, and sample rate reduction before sending the digitized baseband data to the computing
platform via the USB interface. It provides the opposite processing functions for the transmit
path.
Figure 3.1 below shows the high level block diagram of the USRP board with daughterboards
and Figure 3.2 shows an image of the USRP with the basic daughterboards plugged in.
Figure 3.1: USRP and daughterboard block diagram [5]
3.4.1 FX2 USB 2 Controller
The FX2 microcontroller contains an embedded USB 2.0 transceiver and handles all USB
transfers with the upstream USB host. It presents a data bus to the FPGA, with generic control
signals which can be programmed to behave in a custom manner. This interface is called the
General Purpose Interface (GPIF).
The FX2 also handles all USB control requests, which all USB enabled devices must support to
fully comply with the USB standard. These include responses to device capability interrogations
and standard setup requests [32].
3.4.2 ADCs and DACs
The analogue interface portion houses the Analog Devices, AD9862. The AD9862 provides
several functions. Each receive section contains two ADCs. The ADCs operate at 64
Msamples/second with 12 bits of resolution and with reference to the Nyquist criterion the ADCs
are capable of digitising a band 32MHz wide.
Positioned ahead of the ADCs there are programmable gain amplifiers (PGA) with 20dB of gain
available to adjust the input signal level in order to maximise use of the ADCs dynamic range
which evaluates to approximately 72dB of dynamic range. The full range on the ADCs is 2V
peak to peak, and the input is 200 ohms differential. This is equates to 5mW, or 7dBm.
45

Figure 3.2: USRP and daughterboards [39]
The transmit path provides an interpolater and upconverter to match the output sample rate to the
DAC sample rate and convert the baseband input to a low IF output. There are PGAs after the
DACs. The DACs operate at 128 Msamples/second with 14 bits of resolution and can supply a
50 ohm differential load, with 10mW or 10dBm. There is also PGA used after the DAC,
providing up to 20dB gain.
This is shown in the ADC block diagram, in Figure 3.3.
In any digitisation process, the faster that the signal is sampled, results in an effective gain of
received SNR.
46

Furthermore, the relationship between the SNR and sampling rate in the signal bandwidth is
given by:
where:
N
f s
B
SNR = 6.02 N 1.76dB 10 log f s
2B
is the number of bits of resolution of the ADC
is the sampling rate of the ADC
is the Bandwidth of the signal
(3.1)
Figure 3.3: AD9862 ADC block Diagram [38]
47

3.4.3 FPGA
The FPGA employed by the USRP is the Altera Cyclone and performs the high speed signal
processing. Furthermore, the FPGA manages the signal including reducing the data rates to
facilitate their transport over the USB2 interface chip, the Cypress FX2, to the host PC where
less intensive processing takes place. The FPGA and the FX2 chip are programmed over the
USB2 bus.
The clock provided by the USRP drives the ADCs at 64 Msps. If needed the AD9862 may divide
this clock by two to reduce the sample rate. This only affects the clock rate of the ADCs, most of
the sample rate conversion is done in the FPGA and is referred to as decimation in this case.
The block diagram of the USRP board and the high level FPGA signal processing blocks is
shown in the image below.
Figure 3.4: USRP and FPGA block Diagram [39]
Following the signals digitisation, the data is sent to the FPGA. The standard FPGA firmware
provides two Digital Downconverters (DDC), a second FPGA implementation however provides
an additional two DDCs. The FPGA uses a multiplexer to connect the input streams from each of
the ADCs to the inputs of the DDCs.
48

The multiplexer allows the USRP to support both real and complex input signals and allows for
having multiple channels selected out of the same ADC sample stream. The DDCs operate as
real downconverters using the data from one ADC fed into the real channel or as complex DDCs
where the data from one ADC is fed to the real channel and the data from another ADC is fed to
the complex channel via the multiplexer.
The DDC consists of a numerically controlled oscillator (NCO), a digital mixer, and a cascaded
integrator comb (CIC) filter. These components downconvert the desired channel from the IF to
baseband, reduce the sample rate and provide low pass filtering. The effect of the NCO and
multiplier are implemented using the CORDIC algorithm.
This shown in Figure 3.5.
Figure 3.5: DDC block Diagram [39]
Since the USB bus operates at a maximum rate of 480 Mbps the FPGA must reduce the sample
rate in the receive path. The decimator can be treated as a low pass filter followed by a
downsampler. Suppose the decimation factor is D. If we look at the digital spectrum, the low
pass filter selects out the band [−/ D , / D] and then the downsampler spreads the spectrum
to [− ,] . So in fact, we have narrowed the bandwidth of the digital signal of interest by a
factor of D.
The transmit path for the USRP is similar to the receive path, however there are differences.
Since the sample rate the DACs operate at 128 Msps, it is necessary to increase the sample rate
in order to match the sample rates between the high speed data converter and the lower speeds
supported by the USB connection, this process is called interpolation and an interpolater running
on the FPGA increases the data rate. The AD9862 also provides a further data rate increase by a
factor of four. The transmit portion of the AD9862 provides the mixer and NCO required to set
the IF frequency of the transmitted signal, the FPGA performs this function in the receive path.
This is shown in Figure 3.6.
49

Figure 3.6: USRP transmit chain block Diagram [39]
3.4.4 The Daughterboards
On the motherboard there are four slots into which you can plug in up to 2 RX daughterboards
and 2 TX daughterboards. The daughterboards are used to hold the the RF receiver interface or
tuner and the RF transmitter.
There are slots for 2 TX daughterboards, labelled TXA and TXB, and 2 corresponding RX
daughterboards, RXA and RXB. Each daughterboard slot has access to 2 of the 4 high-speed AD
/DA converters (DAC outputs for TX, ADC inputs for RX). This allows each daughterboard
which uses real (not IQ) sampling to have 2 independent RF sections, and 2 antennas (4 total for
the system). If complex IQ sampling is used, each board can support a single RF section, for a
total of 2 for the whole system.
A variety of daughterboards are available, including [6]:
Basic Tx and Rx Boards
The BasicTX and BasicRX are designed for use with external RF frontends as an IF interface.
The ADC inputs and DAC outputs are directly transformer-coupled to SMA connectors 50 Ohm
impedance with no mixers, filters, or amplifiers.
The BasicTX and BasicRX operate in the 1MHz to 250MHz range and give direct access to all
of the signals on the daughterboard interface including 16 bits of high-speed digital I/O, SPI and
I2C buses, and the low-speed ADCs and DACs, and as such are useful for developing custom
daughterboards or FPGA designs.
LFTX and LFRX Boards
The LFTX and LFRX are very similar to the BasicTX and BasicRX, respectively, with 2 main
differences. Because the LFTX and LFRX use differential amplifiers instead of transformers,
their frequency response extends down to DC. The LFTX and LFRX also have 30 MHz low
pass filters for anti aliasing.
50

DBSRX
The DBSRX is a complete receiver system for 800 MHz to 2.4 GHz with a 3-5 dB noise figure.
The DBSRX features a software controllable channel filter which can be made as narrow as
1 MHz, or as wide as 60 MHz.
The DBSRX frequency range covers many bands of interest, including all GPS and Galileo
bands, the 902-928 MHz ISM band, cellular and PCS, the Hydrogen and Hydroxyl radio
astronomy bands, DECT, and others. The DBSRX is MIMO capable, and can power an active
antenna via the coax.
RFX
The RFX family of daughterboards turns a USRP motherboard into a complete RF transceiver
system. An antenna is necessary in order to achieve two-way, high bandwidth communications
in many popular frequency bands. The family of RFX daughterboards cover a frequency range
from 400MHz to 2.9GHz. The boards have many features which facilitate their integration into
more complex systems, such as digital control lines and the option for split transmit and receive
ports.
TVRX
The TVRX daughterboard is a complete VHF and UHF receiver system based on a TV tuner
module. It operates over a 6 MHz wide block of spectrum from anywhere in the 50-860 MHz
range. All tuning and AGC functions can be controlled from software.
Different tuner modules have been used for the TVRx. They differ in the frequency of the IF
output and whether or not they perform a single-conversion or a second down-conversion. In this
dissertation the module using a single-conversion approach to 43.75MHz has been used. The
alternative module offers a second conversion to 5.75MHz.
This board will be used as the front end for the PCL system and thus is used in all the receiver
tests conducted and forms the subject of the receiver characterisation presented in this
dissertation.
Figure 3.7: Signal flow from a single TVRx daughterboard to DDC0
51

The block diagram in Figure 3.7 is representative of the flow path of the TVRx signal. The
daughterboard produces a differential pair as output, shown as the signal S(t) and -S(t). These
signals, after they have been digitised, are routed via the multiplexer to the inputs of the DDC0.
It is at this stage that the complex signal I(n) + jQ(n) is produced. In the instances where the
second TVRx daughterboard is used, a second differential pair is produced, digitised at ADC2
and ADC3 respectively and routed to DDC1 through the multiplexer. Thus two IQ pairs are sent
from the USRP to the PC via the USB for further processing.
3.5 GNURadio
GNURadio is an open source, software defined, radio platform for building and deploying
software radios. It has a large worldwide community of developers and users that have
contributed to a substantial code base and provided many practical applications for the hardware
and software.
It provides a complete development environment to create software radios, handling all of the
hardware interfacing, multi threading and portability issues. GNU Radio has libraries for all
common software radio needs, including;
various modulations:
●
●
●
GMSK
PSK
OFDM
error-correcting codes:
●
●
●
Reed-Solomon
Viterbi
Turbo Codes
signal processing constructs
●
●
●
●
●
optimized filters
FFTs
equalizers
timing recovery
Scheduling
It is a very flexible system, it is user expandable and it allows applications to be developed in
C++ or Python.
3.5.1 The GNURadio Architecture
GNU Radio is organised using a two-tier structure that provides a data flow abstraction. It builds
applications using python code for high level organisation, policy, GUI and other non
performance critical functions, by employing the concept of a graph containing signal processing
blocks and connections for data flow between them.
52

The signal processing blocks are implemented and correspond to some sophisticated functions or
class methods in C++.
Many processing blocks currently exist within the GNURadio framework and include filters,
demodulators and other signal manipulation elements.
A graph can be thought of as a framework upon which all the necessary elements are placed and
then connected. Using a schematic diagram as an analogy for a graph, the elements placed on the
schematic include the resistors, amplifiers, etc. These components are then “wired” up for
current to flow between them. Thus in GNU Radio, the circuit parts are replaced by signal
sources, sinks and the signal processing blocks.
Figure 3.8: GNURadio application structure
Conceptually, blocks process infinite streams of data flowing from their input ports to their
output ports. Blocks’ attributes include the number of input and output ports they have as well as
the type of data that flows through each. The most frequently used types are short, float and
complex. The Simplified Wrapper and Interface Generator (SWIG), is what enables the C++
implementation of the various blocks to be used from Python.
Some blocks have only output ports or input ports. These serve as data sources and sinks in the
graph. The signal source can have various implementations, but is usually the USRP or a file
containing sampled data. The signal stream then flows through any number of processing nodes.
The final output of the processing graph terminates in a signal sink, which may be a real time
spectrum analyser, a real time oscilloscope, an output file, the USRP or audio hardware. It is
estimated that 100 blocks are implemented in GNURadio and new blocks can be written as
desired.
This framework allows for simple implementation of powerful signal processing systems.
53

An elementary application is presented here to present the concepts.
#>>> bring in blocks from the main gnu radio package
from gnuradio import gr
#>>> bring in the audio source/sink
from gnuradio import audio
#>>> create the flow graph
fg = gr.flow_graph()
#>>> create the signal sources
#>>> parameters: samp_rate, type, output freq, amplitude, offset
src = gr.sig_source_f(32000, gr.GR_SIN_WAVE, 350, .5, 0)
#>>> create a signal sink
#>>> parameters: samp_rate
sink = audio.sink(32000)
#>>> connect the source to the sink
fg.connect(src, sink)
#>>> run the flow graph
fg.run()
3.5.2 Common GNURadio Blocks
As mentioned earlier, an estimated 100 signal processing blocks are predefined in the GNU
Radio environment. The best way to become acquainted with these is to explore the GNU Radio
Documentation. This can be found on line at http://gnuradio.org/doc/doxygen/index.html.
For demonstration purposes, two of the frequently used blocks will be mentioned here.
Block: usrp.source_c [s]
Usage:
usrp.source_c (s) (int which_board,
unsigned int decim_rate,
int nchan = 1,
int mux = -1,
int mode = 0 )
The suffix c (complex), or s (short) specifies the data type of the stream from USRP. Most likely
the complex source (I/Q quadrature from the digital down converter (DDC)) would be more
frequently used. which_board specifies which USRP to open, which is probably 0 if there is only
one USRP board. decim_rate tells the digital down converter (DDC) the decimation factor D.
nchan specifies the number of channels, 1, 2 or 4. mux sets input MUX configuration, which
determines which ADC (or constant zero) is connected to each DDC input ‘-1’ means we
preserve the default settings. mode sets the FPGA mode, which we seldom touch.
54

The default value is 0, representing the normal mode. Quite often we only specify the first two
arguments, using the default values for the others.
For example:
usrp_decim = 250
src = usrp.source_c (0, usrp_decim)
Block: calc_dxc_freq ()
Usage:
usrp.calc_dxc_freq(number target_freq,
number baseband_freq,
number fs)
This function is a utility to calculate the frequency to be used for setting the digital up or down
converter. It returns a 2-tuple (ddc_freq, inverted) where ddc_freq is the value for the DDC and
inverted is True if we are operating in an inverted Nyquist zone. target_freq specifies the desired
RF frequency in Hz. baseband_freq is the RF frequency that corresponds to DC in the IF and fs
is the converter sample rate.
3.5.3 Hardware
Various hardware options are available for use with GNU Radio. For instance, the sound card
can be used as an output sink. Wide band I/O and VXI/cPCI cards can also be used as additional
options. However, largely identified as the primary hardware choice as regards GNURadio is the
USRP.
3.6 Summary
Included in the objectives of the overall PCL project is the development of a radar system at
considerably low cost. In pursuit of this, SDR has been identified as a candidate technology. In
this chapter a presentation of the definition and paradigm of SDR is made, including their
applications and the development tools available.
Furthermore, the considerations in the design of SDR receivers is made, facilitating the
characterisation and testing of the SDR toolkit selected for the implementation of our radar
receiver.
In the light of this, the USRP and GNURadio are examined, the USRP provides a flexible and
powerful FPGA, developed by Altera, as the processing unit, which implements a great deal of
the signal processing in the receive and transmit chains. Furthermore, the USRP hosts two
Analog Devices AD9862 chips, each providing a two ADCs sampling at a rate of 64
Msamples/second with 12 bits of resolution and two DACs operating at 128 Msamples/second
with 14 bits of resolution.
55

Moreover, the AD9862 chips provide two auxiliary ADCs and three auxiliary DACs which can
provide further flexibility in computation and control of external daughterboard components.
These daughterboard boards provide front end flexibility allowing for the down-conversion and
thus manipulation of the radio spectrum.
GNURadio, is seen to be a powerful toolkit that implements numerous complex signal
processing elements and additionally interfaces with the USRP. Thus combining to provide a
powerful and widely versatile toolset for the development of a variety of radio devices.
The following chapter tests the USRP and GNURadio in a variety of performance related issues
and therefore, allowing for comment on its usefulness for our application.
56

Chapter 4
Receiver Characterisation
4.1 Introduction
The most critical component of a wireless system often is its receiver. The purpose of this is to
extract the desired signal reliably from the various sources of signals, interference and noise.
The following sections will present the characterisation of the USRP with the TVRx
daughterboard and GNURadio as a receiver in the PCL system being investigated.
At this time, various aspects of the overall project are still under investigation and research. As a
result, the specific requirements that the receiver system must meet are not available. However,
the performance of GNURadio in its control of the USRP with the TVRx daughterboard and
some of the fundamental principles of radio receiver and digital radio receiver design are
identified, discussed and measured, thereby facilitating this examination.
The results of the characterisation will be presented within the body of each section, in order to
facilitate coherency in its reading.
4.2 Test System Overview
This section will describe the manner in which the equipment was setup in order to measure the
various characteristics described in this chapter.
The test signal is generated by a Hewlett Packard 8656B signal generator, with the output
frequency set to 100MHz. The signal generator has the desirable functionality to vary the power
of the signal output generated. The signal is fed into the USRP, setup with the TVRx
daughterboard, and after digitisation is sent to the PC. In addition to this, a Hewlett Packard
54645D oscilloscope is used to probe the pin outs of the TVRx daughterboard. The individual
test conditions will be further described in the relevant subsection.
The test equipment is pictured in Figure 4.1.
57

Figure 4.1: Experimental setup used in characterisation tests
4.3 Receiver Bandwidth
The TVRx daughterboard is built on the Microtune 4937 D15 RF tuner module. The
instantaneous bandwidth by definition is the frequency band over which multiple signals are able
to be simultaneously amplified to within a specified gain tolerance [29]. The bandwidth is an
important property as it affects the noise and signal and has a direct impact on the level of signal
processing gain achievable by the receiver as eluded to by Equation 2.6. This module provides a
static 6MHz bandwidth, referenced to the centre frequency that the module is tuned to. With
respect to our application, the signals of interest are the Broadcast FM signals and the target
returns from the surveillance areas which have a bandwidth of 100kHz. Thus the receiver
provides much more than the requisite bandwidth. Furthermore, we can sustain 32MB/second
across the USB. All samples sent over the USB interface are in 16 bit signed integers in IQ
format, i.e. 16 bit I data followed by 16 bit Q data, resulting in 8M complex samples/second
across the USB. This provides a maximum effective total spectral bandwidth of about 8MHz.
The FPGA provides the functionality of altering the decimation rate and thus allowing us to
select narrower ranges. For instance, the bandwidth of a FM station is 100kHz, by selecting the
decimation factor to be 250, the effective bandwidth across the USB is 64MHz / 250 = 256kHz,
which is well suited for the 100kHz bandwidth without losing any spectral information.
58

4.4 Gain Range and Control
The TVRx daughterboard has two AGC inputs available to level the signal. A block diagram of
this is presented in Figure 4.2. The daughterboard has a RF AGC range of 50dB and an IF AGC
range of 33dB. GNURadio achieves this gain variation by varying the voltage on to each of these
inputs. As mentioned in the previous chapter the ADC that follows the daughterboard provides
an additional level of amplification using its internal PGA, that provides an additional 20dB of
gain.
Figure 4.2: Block diagram showing TVRx gain stages
All three stages are set altogether via the set_gain(gain in dB) method. However, since multiple
signals are to be processed simultaneously, the use of an AGC can result in a reduction of RF
sensitivity and will cause weaker signals to be dropped. GNURadio proves its versatility in this
instance as the set_gain method can be bypassed and each gain stage set manually and
independently.
In the following tests conducted on the receiver and in those of later sections, the AGC was
disabled and the amplifier stages were set independently. Figure 4.3 and Figure 4.4 show the
variation of the voltage applied to the AGC pins of the daughterboards to the gain of the
individual stages.
The RF stage exhibits a linear increase in the gain, which accurately describes the behaviour of
the amplifier between the 10dB and 40dB range. Above 40dB the amplifier begins to run into
saturation [37] and does not continue to show the linear conversion gain. The IF stage
additionally exhibits a linear increase in gain. This is an accurate description of the amplifier
behaviour in the range from 0dB to 15dB. Thus, GNURadio provides us with an accurate
description of the TVRx daughterboard's amplifier behaviour in the region from 10dB to 55dB.
The PGA additionally, has a linear gain curve [38] and provides a further 20dB of gain,
extending the gain range up to 75dB.
59

Figure 4.3: Variation of voltage applied to RF AGC against RF Gain
Figure 4.4: Variation of voltage applied to IF AGC against IF Gain
60

4.5 Multiplexer Usage and Data Interleaving
The FPGA multiplexer is like a router. It determines which ADC is connected to each DDC
input. There are 4 DDCs and each has two inputs. GNURadio controls the multiplexer using the
usrp.set_mux() method.
The multiple RX channels, which can be either 1,2, or 4, must all be the same data rate.
Therefore, the same decimation ratio must apply to all the channels. This condition additionally
applies to the TX channels, however they may be different to the RX rate. When there are
multiple channels, GNURadio interleaves the channels. This means that with 4 channels, for
instance, the sequence sent over the USB would be:
←
DDC0 DDC1 DDC2 DDC3 DDC0 DDC1 DDC2
I 0 Q 0 I 1 Q 1 I 2 Q 2 I 3 Q 3 I 0 Q 0 I 1 Q 1 I 2 Q 2
←
The USRP can operate in full duplex mode, the transmit and receive sides are completely
independent of one another. The only consideration is that the combined data rate over the USB
must be 32 MB/sec or less.
The set_mux parameter is 32 bits, each nybble of 4 bits controls which ADC is connected to
which DDC input. The least significant nybble of the parameter represents DDC0 and the most
significant nybble of the parameter represents DDC3.
DDC3 DDC2 DDC1 DDC0
Q 3 I 3 Q 2 I 2 Q 1 I 1 Q 0 I 0
Figure 4.5: Multiplexer setup for two TVRx to DDC0 and DDC1
61

In Figure 4.5, the multiplexer parameter is set as Ox----3210, we are not concerned about the
upper nybble as those bits set DDC3 and 2, represented here by dashes. However, we see that the
I and Q values of DDC0 and DDC1 are set up accordingly.
4.6 Frequency Range and Control
The TVRx daughterboard that we are using uses a single conversion approach to 43.75MHz with
the reception frequency divided into the VHF low, VHF high and UHF bands from 50MHz to
860MHz. The band selection and tuning is done via the I 2 C bus, this coordinated from within
GNURadio. The tuning range is the frequency band over which the receiver will operate without
degrading the specified performance [29].
The oscillator frequency is driven by a 4MHz crystal reference and determined by the following:
f osc
= f ref
× 8 × SF
(4.1)
where:
f ref
SF
is the crystal reference frequency / Reference divider
is the programmable scaling factor
The reference divider mentioned above can take on one of three values, those being 512, 640 and
1024. This value is set from within GNURadio and the default is set at 640. The value of this
reference divider influences the tuning step size, otherwise understood as the granularity of the
tuner in setting the local oscillator frequency. The default choice results in a tuning step size of
50.0kHz with a potential minimum of 31.25kHz. This level of resolution is adequate for our
application.
GNURadio sets up the appropriate byte values in the I 2 C data by use of the following method.
Block: usrp.tune ()
Usage:
usrp.source_x.tune (u,
chan,
subdev,
target_freq)
Tuning is a two step process. First we ask the frontend to tune as close to the desired frequency
as it can. Then we use the result of that operation and our target_frequency to determine the
value for the digital down converter. The function returns False if there is a failure, in the case of
success it returns tune_result. U is the instance of the usrp that we want to tune. chan is the
DDC channel that we wish to tune. Subdev tells the method which daughterboard subdevice to
tune and target_freq is the centre frequency in Hz.
62

However, GNURadio does offer the flexibility to control individual parts of the tuning process,
e.g. tuning the TVRx frontend and setting the value of the digital down converter independently.
4.7 Spurious Artifact
During testing it was found that when a GNURadio application is run, the USRP generates a
large spurious output, this applies additionally to instances where the application is stopped and
started again. This spurious output does corrupt data that is being stored to memory on the PC
for offline post processing. However, when the USRP is used for real time processing it does not
present a concern, as the spurious output occurs at the start of the data. Figure 4.6 shows this
artifact at the beginning of a data.
Figure 4.6: Spurious signal generated by USRP on startup
Figure 4.7 shows a zoomed in image of this artifact, illustrating the corruption of the first
approximately 80 samples. In order to mitigate the effects of this artifact, this data is simply cut
off and the number of remaining samples made note of in the processing that follows.
63

Figure 4.7: Closer look at the spurious signal generated by USRP on startup
4.8 Receiver Noise Figure
The noise figure is a measure of how much noise a component or system adds to a signal. The
noise figure of a system depends on the losses in the circuit, the nature of the solid state devices,
any bias applied and on amplification, to name but a few. When noise and a desired signal are
applied to the input of a noiseless system, it is expected then that the noise and signal would be
amplified or attenuated by the same factor, thus the SNR through the system would remain
unchanged.
A noisy system, however, will introduce noise. Thus the noise power will increase relative to the
signal power, thereby reducing the SNR at the output of the system.
The noise figure of a component is, by definition, independent of the noise input into the
component and thus is based on a standard input noise source N i , at room temperature in a
bandwidth B , thus:
N i
= T 0
B
(4.2)
Where:
T 0
is the Boltzmann constant
is the room temperature 290K
64

Assuming a 1Hz bandwidth this value evaluates to the standard, 4x10 -21 W, which can be further
expressed as -204dBW, which is further still expressed as -174dBm. The noise figure in essence
is the difference between the noise output of the receiver and the noise output of an ideal,
noiseless receiver.
Noise figure is the decibel notation of the noise factor, which is defined as:
F =
SNR at input
SNR at output = S i /N i
S o
/N o
(4.3)
In a typical receiver, the input signal will pass through a number of cascaded components such
as filters, amplifiers and mixers, among others. Each stage will inject a degree of noise into the
signal, degrading the signal to noise ratio. This makes it important to quantify the overall effect
of the noise figure on the system's performance.
Consider the network shown in the figure below, the total noise factor referenced to the input to
the system is determined using the Friis equation as:
F total
= F 1
F 2 − 1
F 3 − 1
F N
− 1
⋯
G 1
G 1
G 2
G 1
G 2
⋯G N − 1
(4.4)
Figure 4.8: Cascaded components
The characteristics of the cascaded system are dominated by the earlier stages. This is as the later
stages are reduced by the product of the gains of the preceding stages. Thus the ideal scenario is
one in which the first stage or component of a system has a low noise figure and a high gain to
ensure the low noise figure for the overall system.
In order to determine the noise figure of the receiver system, a known signal of low magnitude is
input. In this instance -50dBm was input. The USRP was setup with a decimation rate of 64, thus
providing an effective 1MHz of bandwidth for analysis. A spectrum analyser was setup, with
1024 points of measurement resolution thus providing a reading representing 1kHz per
measurement bin. This setup allows for the measurement of the noise floor per bin, by observing
how far down the noise is from the input signal. The noise floor is a further 38dB below this
level, measured in dBm/Hz. The noise figure is then determined by the difference between this
signal and the theoretical level of -174dBm/Hz, derived from Equation 4.2. The noise figure of
the TVRx and the USRP is thus determined to be 10dB.
65

4.9 Sensitivity
The sensitivity of a receiver is defined as the lowest power level at which the receiver can detect
an RF signal and demodulate data; it is a specification that is associated solely with the receiver
system and is independent of the transmitter. As the signal propagates away from the transmitter,
the power density of the signal decreases, this makes it more difficult for a receiver to detect the
signal as the distance increases. Receivers with desirable sensitivity levels are able to increase
the range over which surveillance or reception of meaningful signals is possible.
The more common methods of specifying the sensitivity of a radio receiver include using SNR
or Signal to Noise and Distortion ratio (SINAD). In the case of the SNR method, the sensitivity
is realised by determining the input signal power necessary in order to achieve a specified SNR.
The predetermined SNR is a function of the receiver application. For the experiments conducted
here the SNR is set to be 10dB.
It is worthwhile to note that, in addition to the basic performance of the receiver, the receiver's
bandwidth can affect the SNR specification. It is found that the noise power decreases as the
system bandwidth decreases. This implies that systems with smaller bandwidths collect less
noise power [23].
The SINAD measurement takes into account the degradation of the signal by unwanted
contributions including noise and distortion. More specifically, SINAD is defined as the ratio of
the total signal power level, that being the signal, the noise and the distortion to the unwanted
signal power, that being the noise and distortion.
SINAD = 10log SND
ND
(4.5)
where:
SND
ND
is the combined signal, noise and distortion power level
is the noise and distortion power level
Similarly the sensitivity is assessed by determining the input level required in order to achieve a
given figure of SINAD.
In Figure 4.9, it is shown how the sensitivity of the TVRx and USRP varies. As the level of gain
is increased, it is shown that a minimum signal of -105dBm is needed in order to produce the
requisite 10dB of SNR.
66

Figure 4.9: Variation of receiver sensitivity
4.10 Dynamic Range, Minimum Detectable Signal and
Gain Compression
The standard operation of a receiver system is typically in the region where the output power is
linearly proportional to the input power [3]. The constant of proportionality in this region is
known as the conversion gain or conversion loss. It is this region of proportionality that is
referred to as the dynamic range of a receiver. If the input power goes below the minimum
acceptable signal, the minimum detectable signal (MDS), the noise effects dominate the receiver
and results in nonlinear behaviour. This is due to the effects of the various sources of noise into
the receiver including the components themselves.
If the input power goes above the maximum allowable power of this region the receiver will
display nonlinear characteristics as well. This behaviour may be as a result of damage to the
receiver components at high power levels or due to gain compression or the generation of
spurious frequency components, due to device nonlinearities. These effects are undesirable as
they may lead to increased losses, signal distortion and depending on the application possible
interference with additional radio channels or services.
67

Thus the definition of dynamic range can be summarised as:
Maximum allowable signal power (dB) - Minimum detectable signal power (dB)
In the first instance, the maximum allowable signal has been shown to be as a result of gain
compression or saturation. Physically this effect is due to the limitation on the instantaneous
power output imposed by the power supply used to bias the device [23]. In order to quantify the
range over which the device is considered to be operating in the linear region, the 1dB
compression point is defined. This is, the power level for which the output power of the device
has deviated by 1dB from the ideal characteristic, for receiver systems this is usually specified in
terms of the input power.
Although 1dB compression points are most commonly used, 3dB and 10dB compression points
are also used in some system specifications. The 1dB compression point is an important measure
which can be used in the derivation of a receiver system's gain, bandwidth, noise figure and
dynamic range [3]. In this instance, dynamic range is defined as: DR = P in, 1dB
− MDS .
This is shown in Figure 4.10.
Figure 4.10: MDS, 1 dB compression point and dynamic range [3]
The maximum allowable signal has been shown furthermore to be as a result of the generation of
spurious frequency components. This leads to an alternative definition of dynamic range, the
spurious free dynamic range (SFDR). When a single frequency or tone is considered, generally
68

the output will consist of harmonics of the frequency in the form n 0
Usually these harmonics do not affect the desired signal frequency.
, where n = 0,1,2,
This, however, is not the case when the signal is comprised of two or more closely spaced
frequencies 1 and 2 . In particular, the third order intermodulation products,
2 1
− 2 and 2 2
− 1 , are of interest. These terms are located near to the input
frequencies and generally they are of relatively large magnitude, making them difficult to filter
from the desired output and resulting in distortion of the output signal.
Thus the third order intercept point (IP3 or TOI) is defined, as a measure of the intermodulation
suppression. A high intercept point is indicative of a high suppression of undesired
intermodulation products and is the fictitious point where the desired signal power and the third
order signal powers are equal. In general the IP3 point occurs at about a 12dB - 15dB higher
power level than the 1dB compression point. The IP3 point is an important measure of a systems
linearity and is additionally specified in terms of the input power to the receiver.
In this instance, dynamic range is defined as: DR sf
= 2 3 IP3 − MDS .
This is shown in Figure 4.11.
Figure 4.11: Spurious Free Dynamic Range [3]
69

Figure 4.12: Input power against output power, showing receiver dynamic range
Figure 4.12 shows the variation of the output power with the power input to the receiver; from
this we see as gain is increased from 0dB the MDS is determined at increasingly lower power
levels. This is indicative of the increasing receiver sensitivity. Furthermore, as the gain setting
increases we see three distinctive bands as each of the amplifier stages of the TVRx begin to run
into saturation, thus the 1dB compression point begins to travel lower. This effectively indicates
the reduction of the receiver's dynamic range at these high gain levels. The best achievable
dynamic range of the receiver is noted as 62dB. This is in the region, mentioned in section 4.4,
where GNURadio most accurately modelled the TVRx amplifier stages.
Figure 4.13 relates the measured dynamic range of 62dB to the theoretical SNR of the 12 bit
ADC, which is calculated to be approximately 72dB using Equation 3.1. In practice however, the
dynamic range available from the ADC is reduced as a result of thermal noise, reference noise,
clock jitter, quantisation noise, etc. Figure 4.13 shows a 6dB loss in the ADC dynamic range as a
result of noise, thus providing an effective 66dB of dynamic range. The minimum detectable
signal from the TVRx, using the gain setting determined from Figure 4.12, was found to be 3dB
above the ADC noise floor. As a result a further 3dB is lost from the ADC dynamic range as
weaker signals from the TVRx will be noisy and not useful. Furthermore, the 1dB compression
power output from the TVRx is found to be 1dB within the range that the ADC can extend to.
Although this 1dB is available to the TVRx, stronger signals into the TVRx will begin to drive
the amplifiers into saturation. Therefore, although 66dB of dynamic range is available at the
ADC, the TVRx is able to provide 62dB of dynamic range.
70

Figure 4.13: TVRx and ADC dynamic range comparison
4.11 Summary
A number of the TVRx and USRP receiver system parameters as well as the GNURadio
interface to the controllable aspects are examined. The receiver's frontend provides a static
bandwidth of 6MHz and a tunable range between 50MHz and 800MHz, with a tuning step size
as low as 31.25kHz. This is sufficient for the manipulation of FM Broadcast signals as the radar
waveform and provides further functionality and versatility to incorporate additional areas of the
frequency spectrum if desired, i.e. the television video and sound carriers.
The multiplexer implemented in the FPGA and the ability to directly control its operation
through GNURadio further extends the receiver's flexibility in its digital manipulation of signals
through it. The noise characterisation of the receiver reveals a NF of 10dB, a sensitivity of
-105dB and a dynamic range of 62dB.
71

Furthermore, it is observed that when the GNURadio application is started, when the USRP is
setup as a source and includes the occasion when the application is stopped and restarted, a
spurious spike occurs at the beginning of the data set. This phenomena is unexplained but noted
by the USRP developers, The effects of which are mitigated by simply cutting out that part of the
data set.
The following chapter will examine the suitability of FM broadcast signals as a candidate for the
waveform or part of a group of waveforms used in the PCL radar. This is achieved by
examination of the ambiguity plot and the variation of the instantaneous signal bandwidth.
72

Chapter 5
Ambiguity Plot Analysis
5.1 Introduction
In considering the use of broadcast FM signals for our PCL system, it is necessary to fully
understand the nature of the signal parameters and the resulting effect and limitation on the
system's performance. Range and doppler resolution in addition to range and doppler ambiguity
are such parameters and they govern the ability to distinguish between two or more targets by
virtue of their spatial or frequency differences and the ambiguity function has long been used to
evaluate them [29] [24] [22].
The nature of the transmitted waveform determines these properties, for example, FM
broadcasting uses the frequency range 88 – 108MHz [21] and the specified bandwidth is
approximately 150kHz which gives a range resolution of 1–2km. This is further exemplified in
the consideration of the nature of analogue television whereby the 64µs line repetition rate
generates strong ambiguities [12] and consequently imposes a severe restriction and
complication on its performance and applications.
As can be imagined, the programme content and thus the instantaneous modulation of a
particular broadcast FM signal varies with time. Therefore, the ambiguity behaviour will vary as
a function of the time and it is necessary to determine this behaviour.
5.2 System Overview
The receiving system used to capture and digitise the FM broadcast signals is described here.
The system was situated on level 6 of the Menzies building at the University of Cape Town and
detects and digitises broadcast signals originating from the Constantiaburg transmitter. It uses an
antenna feeding the USRP, with the TVRx daughterboard plugged in and is thus in line with the
project's low budget objectives. The daughterboard provides the down-conversion of the FM
signal to the IF of 43.75MHz and is subsequently digitised on the USRP. The digitised signal is
then further down converted to baseband and then decimated, using the standard USRP FPGA
73

implementation. The data stream is then transported via USB and stored in the memory of a
standard PC.
Figure 5.1: Experiment setup to capture signals for ambiguity function analysis
This choice of setup allows for the system to be flexible, as the TVRx module can be tuned to
operate over the frequency range of interest and the captured data can be processed off-line, as
desired.
We then compute the ambiguity function and the results of this computation are presented in a
later section. Figure 5.1 above shows the experimental equipment and Figure 5.2 below
represents this schematically.
Figure 5.2: Block Diagram of experimental setup
74

5.3 Algorithm Concept
The purpose of this investigation is to determine the best achievable resolutions offered by the
FM broadcast signals and the variation of these signal properties with respect to time. It is
important to understand this variation due to its effect on the overall range and doppler
resolutions. For this process, knowledge of the relative positions of the transmitter, receiver and
target are not required and thus the monostatic geometry is assumed in the following setup. This
work is based upon the literature presented by Baker et al and by Howland et al and duplicates
their research.
As previously discussed, the nature of the transmitted waveform determines these properties and
is evaluated by computation of the ambiguity function, given by Equation 2.21.
The output of a matched filter is represented by the ambiguity function and thus, computation of
the ambiguity function results in a three dimensional plot for which one axis is time delay, the
second axis is Doppler frequency and the third is the normalised output power, computed by
matched filtering the directly received transmitter signal. This processing is exemplified in the
figure below.
Figure 5.3: Signal processing algorithm[16]
In the context of a PCL system, this signal is known to be the reference signal and is the
waveform that is correlated with the indirect target scattering in order to produce the output that
is applied to the CFAR detectors, to obtain range and doppler information of each target.
This processing is written in discrete time notation as:
∣ R R , f d ∣ 2 =∣
N ∑ −1
n=0
2
sns ∗ n−R R
exp[ j 2 f d
n/ N ]∣
(4.6)
75

A 0.1 second data sample of each FM Broadcast frequency of interest is digitised, with the
USRP decimation rate set to the maximum value (256) this reduces the processing load and
lowers the algorithm computation times. The effective sample rate is thus set to be 250kHz,
which is well within the Nyquist criterion. The algorithm operating on each sample of data
generates doppler shifted copies of the reference signal, each matched to a different target
velocity.
The implementation of the processing begins by rotating and then conjugating the samples of the
reference signal sn , producing the values representing the time delay, s ∗ n−R R
.
These values are multiplied with the original reference signal and produce the result
sn s ∗ n−R R
. The Fourier transform of this result is determined and repeated for each
range of interest.
The USRP and GNURadio were setup to digitise twenty successive 0.1 second samples, thereby
providing 2 seconds of contiguous data for the analysis of the variation of range resolution with
respect to time, for the transmission types of interest. The findings of this analysis follow.
5.4 Results
The ambiguity plots derived from the system above will be analysed in this section, illustrating
the variability of the responses. Only the first set of plots and the associated table illustrating the
variation of the signal bandwidth will be shown in this section.
The ambiguity plots and the associated zero doppler and delay cuts as well as the tables of the
remaining signals will be presented in appendix A, however a discussion of the results will be
included in this section.
The ambiguity plots shown are those of the six FM radio stations available from the
Constantiaburg transmitter, namely these are (a) RAD5, the content on this channel is comprised
mainly of rock, pop, RnB,hip hop and dance. This is dependant on the time of the day and week.
Every hour, however, a five minute news bulletin is broadcast. (b) Radio 2000, the majority of
this content is a mixture of rock, jazz and talk. (c) Umhlobo, depending on the time of day
broadcasts talk shows in the Xhosa medium, RnB and reggae. (d) Classic FM, which broadcasts
an array of contemporary and classic works. (e) RGHP, offers a selection of rock and country
music. (f) RSGR, the dominant medium of the content that is broadcast is Afrikaans. Talk
shows, pop, rock and country are all broadcast. (g) SAFM, which broadcasts news and talk
programmes of various subjects and topic.
Figure 5.4 below shows the ambiguity function for a RAD5 transmission. The signal content was
comprised of rock. The ambiguity function plot shows a narrow defined peak and demonstrates
fast fluctuating detail in the sidelobe regions. This indicates a wide bandwidth and due to the
randomness in the signal modulation, exhibits noise-like characteristics and behaviour. which is
consistent with what might be expected for rock music as compared to speech, which will be
shown later. The zero range and doppler sections, in Figure 5.4a and Figure 5.4b below, show
sidelobes to be usefully low with good performance in the range domain, which is reflective of
the high rate modulation in the rock waveform. Sidelobe levels are good with about 27dB in the
frequency domain and around 33dB for range.
76

Figure 5.4: Ambiguity plot of Rad5 transmission
Figure 5.4a: Zero doppler cut through ambiguity plot
77

Figure 5.4b: Zero delay cut through ambiguity plot
Upon analysis of the time dependant performance of this transmission, over a 2 second period,
the behaviour is shown to be relatively consistent. This is illustrated in Table 5.1 below, which
shows the bandwidth in kHz for twenty waveform samples.
Table 5.1: Bandwidth Variation of RAD5 waveform
Elapsed Time (s)
Bandwidth (kHz)
0.1 30
0.2 50
0.3 42
0.4 55
0.5 45
0.6 52
0.7 40
0.8 43
0.9 10
1.0 32
1.1 50
1.2 47
1.3 40
78

1.4 40
1.5 44
1.6 50
1.7 47
1.8 43
1.9 40
2.0 42
Average Bandwidth (kHz) 44.38
The bandwidth is seen to vary in between 30kHz and 55 kHz, however, an anomaly of 10kHz is
observed. This can be attributed to a break or pause in the modulating signal. The signal
bandwidth averages 44.38kHz, making use of approximately 30% of the available bandwidth.
Consequently, as the bandwidth is shown to be a function of time, the performance of the radar
system will also be a function of time.
Figure A.1 shows the ambiguity function for a Radio 2000 transmission.
The ambiguity function plot again shows a defined peak, indicating a waveform with noise like
properties. In this instance, distinct peaks in the sidelobe regions are observed, and thus
potential ambiguity. However, the zero range and doppler sections provide further insight and
show the sidelobe levels are good with about 30dB in the frequency domain and around 30dB for
range and useful resolution in range and thus eliminating concerns over ambiguity.
Table A.1 examines the time dependent bandwidth of the Radio 2000 transmission. Generally
the values of bandwidth are a factor of two lower than that of the RAD5 transmission, this can be
attributed to the nature of the modulating content. The variation swings between 0.5kHz to
50kHz, which is much greater than that shown above and much more significant in its influence
of the radar's performance. Additionally, the modulation bandwidth measured is 13% of that
available.
Figure A.2 , derived from a Classic FM transmission, shows a well defined ambiguity function
peak, however,wider than those we have seen thus far, this can be attributed to the narrower
bandwidth associated with this waveform modulation.
The zero range and doppler sections shown in Figures A.2a and A.2b demonstrate the range and
doppler resolutions more clearly. The range resolution shows a slight degradation in
performance, as is expected, due to the narrower bandwidth offered. Sidelobes are seen close to
the main peak but with reasonable levels at about 26dB in the range domain and around 40dB in
the frequency domain. The peaks seen in the frequency domain away from the peak is measured
at 33dB and thus does not impede the doppler performance.
Table A.2 exhibits a variation in bandwidth from 0.6kHz to 32.5 kHz of the Classic FM
transmission. The signal bandwidth averages 18.81kHz, making use of approximately 13% of
the available bandwidth.
Figure A.3 shows the ambiguity function for a Umhlobo FM transmission.
The ambiguity function plot again shows a narrow defined peak, offering attractive range
resolution and some detail in the sidelobe regions. In this analysis the signal content was
comprised of kwaito music. This is loosely described as a South African variant of rock.
Examining the zero range and doppler sections exhibit fast fluctuating detail associated with
signals modulated with pop music. Indicating a more noise like waveform structure and a wider
bandwidth than that associated with rock, jazz and classic music. Doppler sidelobe levels are
79

about 39dB and range sidelobe levels are additionally attractive with around 28dB, the range
resolution offered by the signal is useful.
Table A.3 examines the time dependent bandwidth of the Umhlobo FM transmission. Generally
the values of the bandwidth offered are greater and the variation measured is less. The
modulation bandwidth measured is on average 30% of that available.
Figure A.4 shows the ambiguity function for a RGHP transmission.
The ambiguity function plot again shows a defined peak, although narrower, relative to the other
plots examined. This is indicative of the greater bandwidth offered by the modulating music
waveform, in this instance RnB music. In the regions away from the main peak, we see a very
clean structure with attractive sidelobe behaviour. This behaviour is more reflective of a random
noise-like behaviour, which approaches the idealised “thumbtack” surface. An examination of
the zero range and doppler sections displays fast fluctuating detail and is associated with signals
modulating music. Indicating a more random and noise like waveform structure and a wider
bandwidth than those examined thus far. Doppler sidelobe levels are about 32dB, which shows
some measure of degradation relative to the previous signals, although not significant for our
application. Range sidelobe levels however are good with around 28dB, with a very good range
resolution offered.
Table A.4 examines the RGHP transmission. Generally the values of bandwidth are much
improved over the alternative modulating types, averaging 63.6kHz, which translates to 42% of
that available.
The signal from RSGR, in Figure A.5, was comprised of Afrikaans music. The peak of the
ambiguity function is well defined but wider than those we have seen thus far for the alternative
modulation types. The detail in the regions away from the main peak does not fluctuate as fast as
the alternative signals examined thus far, this too is a function of the modulation present in this
component of the waveform. This behaviour is not reflective of pure noise-like behaviour,
however, is consistent with the matched filter response that might be expected with a narrower
bandwidth. The zero range and doppler sections shown below in Figures A.5a and A.5b
demonstrate the range and doppler resolutions more clearly. The range resolution shows a
degradation in performance, consistent with the decrease in the bandwidth associated with the
signal. Sidelobe levels however are good with about 36dB in the frequency domain and around
30dB for range.
Table A.5 below demonstrates the behaviour of the RSGR transmission. Generally the values of
bandwidth are less than that of the rock, pop, and classic music modulating signals. The
bandwidth is on average 11% of that available.
The signal from SAFM in Figure A.6 was comprised of speech in a news broadcast. The peak of
the ambiguity function is extremely wide. This too is a function of the modulation present in this
component of the waveform and is consistent with the correlation that might be expected of a
pause or break in speech. The zero range and doppler sections shown in Figures A.6a and A.6b
demonstrate the range and doppler resolutions and sidelobe levels with about 40dB in the
frequency domain and performing extremely poorly in range. This result is indicative of
ambiguous performance as the bandwidth offered is very low, thus the potential resolution of the
radar would be not at all useful in this instant.
In Table A.6, the bandwidth is seen to vary from 1.4kHz to 47.5kHz. This average bandwidth
used is 13% of that available. This is consistent with the analysis expected of a speech (news
broadcast) transmission.
80

All the bandwidths presented here are calculated from the -3dB width of the zero doppler
section, through the ambiguity plot. It is evident that there is a great deal of variation in the
performance of the available waveforms to be exploited. This variation is dependent on the type
of transmission, its content and the time of transmission. This can be understood, for example, if
there is a long pause on a channel or with speech, the signal spectral content is reduced, thus
range resolution information will be degraded or lost.
This is illustrated in Figure 5.5a and Figure 5.5b, which compares the range resolution against
time for the different transmissions.
Figure 5.5a: Variation of signal range resolution with time
81

Figure 5.5b: Variation of signal range resolution with time
Here we see that RSGR and SAFM (talk show) show a high degree of temporal variability in
range resolution compared to the music channels as is expected, with the resolution degrading in
the order of hundreds of kilometres. The classical music does not show as much variation, thus
demonstrating an increased performance. However, the pop and dance channels exhibit the least
variation, with resolution varying between 1km and 15km.
Thus the signal performance has quite a considerable variation and the effect on overall system
performance will require careful consideration.
5.5 Summary
From the results it is evident that the measured FM Broadcast signals due to the randomness in
their modulation, exhibit noise like characteristics and behaviour. This is observed as the
ambiguity function plots can be approximated by the ideal thumbtack ambiguity function,
therefore providing the radar with excellent range and doppler information. However, it is
additionally clear that the ambiguity function depends largely on the modulating format.
82

Thus there is a variation in performance between the pop music and dance music channels and
the rock and classical music channels which exhibit a slightly degraded performance. Speech
content shows the poorest performance though, as the performance does vary significantly with
time, especially during pauses in words due to the low spectral content. This variation in
performance does have a significant impact on the radar's potential detection ability and will be
one that does vary with time, as seen from Equation 2.6 and Table A.6, this variation translates
to a change in processing gain of 15dB.
Table 5.2 below summarises the ambiguity function performance of the measured signals.
Signal
Table 5.2: Summary of waveform ambiguity function performances
Range Resolution
(km)
Effective
Bandwidth
(kHz)
Peak range
Sidelobe level
(dB)
Peak doppler
Sidelobe level
(dB)
RAD5 3.38 44.3 - 30 - 27
Radio 2000 7.94 18.9 - 18 - 34
Classic FM 7.98 18.8 - 30 - 33
Umhlobe 3.33 45.0 - 31 - 25
RGHP 2.36 63.6 - 30 - 37
RSGR 9.32 16.1 - 18 - 36
SAFM 7.54 19.9 - 20 - 40
Average 5.98 32.37 -25.29 -33.14
83

Chapter 6
Oscillator Stability
6.1 Introduction
A significant limiting factor in the performance of radar applications is the ability of the radar's
frequency reference to maintain timing accuracy.
Reviewing the theory of previous chapters, it is understood that in radar, the range of the target is
related to time.
R = c⋅ d
2
Where:
R
d
is the range of the target, from the radar receiver.
is the round trip time delay of the signal, between the transmitter and receiver.
Thus, a radar’s range accuracy is directly proportional to the error in the timing signal [24] , as
an error in d is directly proportional to an error in R .
In a doppler radar, the signal received by the radar from a moving target differs in frequency
from the transmitted frequency and thus differs in phase too, by an amount that is proportional to
the radial component of the velocity relative to the radar.
This shift is is given by:
f d
= 2v r
As a result, the velocity of the target and the radar frequency are primary determinants of the
phase noise requirements [27]. A fast moving target will cause a large doppler frequency, and
thus will require a low phase noise at frequencies far from the carrier. In quartz crystals most of
the noise power lies close to the carrier [3] [23], consequently, low phase noise close to the
carrier, required for a slow moving target, is not easily achieved [24].
84

Moreover, coherence, in radar applications is defined as whether the phase relationship (at the
carrier frequency) between successive pulses are known and stable [24]. If a radar is coherent,
then M successive pulses, from a non-fluctuating target, can be added. This summing will
improve the radar's SNR by a factor of M. However, if the receiver is not coherent, these pulses
will not add in phase and summing will not improve the SNR. Refer to chapter 2.6.4 for a more
thorough treatment of this point.
Furthermore, as the radar's probability of detection is directly related to the SNR, degradation of
the radar's coherence will decrease the SNR. As a result, in coherent radar and thus PCL, the
target detection capability decreases with an increase in phase noise of the frequency reference.
In a bistatic or multistatic radar configuration, the complexity of synchronisation and coherence
is further increased. The local oscillators in all the receivers, need to be synchronised not only in
frequency but the phase offsets between them needs to be known as well [27]. The degree to
which these phase offsets stay constant during target detection will influence the level of
coherence. In a passive bistatic radar system the echo signal is correlated with a bank of doppler
shifted replicas of the transmitted signal, over a predetermined sample period, in order to achieve
the necessary signal processing gain and for target detection. Thus the phase coherence is only
required during this sample period. The degree of coherence can be thought of as how stable the
local oscillator clock, in the TVRx, remains.
Assuming a target velocity of 150m/s, a broadcast FM transmission frequency of 100MHz and
using Equation 2.17, a target will produce a doppler shift of 100Hz. In order to qualify the results
of this chapter a 5% error in the calculated doppler shift is set as the maximum acceptable limit
of frequency shift. Thus a frequency drift of no greater than 5Hz is set as the requirement of the
system.
6.2 Quartz Crystal Oscillators
The local oscillator in the TVRx daughterboard of the USRP is derived from a 4MHz quartz
crystal oscillator. For this reason discussion will focus on the properties of crystal oscillators and
the associated performance parameters of interest. These parameters of interest include accuracy,
reproducibility and stability, although, oscillator stability will dominate the discussion.
Accuracy, in the case of frequency, is a measure of how well the frequency source relates to the
definition of one second. Reproducibility, is a measure of how well a number of sources agree in
frequency as they are adjusted, this is more relevant as a factory acceptance test. Stability is a
measure of how well a frequency source is able to generate a given frequency over some
measure of time, once it has been set. Thus the difference between the frequency at one moment
in time and another moment is called stability [19] and is usually given for a number of time
periods, ranging between seconds and years.
Crystal oscillators are widely used as frequency sources and find application in a wide variety of
areas from wristwatches to complex and elaborate instruments found in laboratories. These
oscillators provide good performance at a reasonable price and dominate the field of frequency
sources [19].
The quartz crystal in the oscillator resonates mechanically and the associated oscillations of the
resonator have to be sensed. This is done by taking advantage of the piezoelectric effect. The
piezoelectric effect is observed when a physical compression of the crystal generates a voltage
across the crystal. Conversely, the application of an external voltage across the crystal causes the
crystal to either expand or contract depending on the polarity of the voltage.
85

Limitations in stability stem mainly from oscillator drift and ageing effects. Further discussion
regarding this will follow in section 6.2.1.
6.2.1 Ageing and Temperature of Crystals
The most influential factors affecting oscillator performance are ageing and temperature. There
is a temperature dependence of the quartz crystal that affects its resonance frequency, and there
is also instability of the resonance frequency due to ageing [19]. Drift is due to ageing plus
changes in the environment and other factors external to the oscillator, thus oscillator drift is
defined as the systematic change in frequency with time of an oscillator. Ageing, on the other
hand, is defined as the systematic change in frequency due to physical changes in the oscillator
[27]. This ageing rate decreases with time.
Ageing is a characteristic common to crystal oscillators and can be approximated as a linear
change in resonance frequency with time. In general, the drift is negative, meaning that the
resonance frequency decreases. This decrease is indicative of an increase in the size of the
crystal. Possible causes of this phenomenon are contamination on the surface of the crystal,
variations in the electrodes or the metallic plating, the movement of loose surface material from
grinding and etching or changes in the internal crystal structure.
Furthermore, the continuous expansion and contraction of the crystal, due to the piezoelectric
effect, may be the cause of or at least exacerbating these effects. Improvements in crystal holder
design combined with clean vacuum enclosures, have led to a reduction in oscillator ageing [19].
The temperature effects on crystal performance are as a result of slight changes in the elastic
properties of the crystal. Special crystal cutting techniques and variations known as
crystallographic orientations minimize this effect of temperature over a range of temperatures.
Temperature coefficients of less than one part in 100 million per degree of temperature are
possible.
Crystal oscillators must be carefully designed if very high frequency stabilities are desired. If
large environmental temperature fluctuations are to be tolerated, the crystals themselves are
enclosed in electronically regulated ovens which maintain a constant temperature. These
oscillators are known as Oven Controlled Crystal Oscillators (OCXO).
6.3 Characterisation of stability
The techniques associated with the specification of stability are presented in the following
sections and in order to appreciate this, a presentation of the relationship between frequency and
phase will, additionally, be made.
It is assumed that the average output frequency of a precision oscillator is determined by a
narrow band circuit (crystal oscillator), so that the signal can be approximated as a sine wave
[19] and thus the signal output is generalised by the following formula.
V t = [V 0
t ]sin [2 v 0
t t ]
(6.1)
86

where:
V 0
t
v 0
t
is the nominal peak voltage amplitude
is the deviation of the amplitude from the nominal
is the nominal fundamental frequency
is the deviation of the phase from the nominal
This can be furthermore generalised and rewritten as the t in precision oscillators is
generally ignored.
V t = V 0
sin [2v 0
t t]
= V 0
sin t
(6.2)
where:
t
is the total oscillator phase
Here we note that due to a change in amplitude, v , over an associated change in time, t ,
we able to determine the average frequency over this period. Furthermore, in the limit as t
approaches zero an instantaneous frequency can be measured. This is however, in practice, not
possible as it requires an infinite bandwidth. As a result we are only able to measure the
frequency that has been averaged over a time period t .
Furthermore, the frequency of a signal is additionally related to the rate of change of its phase.
The instantaneous angular frequency is defined as the time derivative of the total oscillator
phase. Thus,
t = d dt [2v 0 t t]
(6.3)
= 2 v 0
d
dt
The instantaneous frequency is then written as:
V t = v 0
1
2
d
dt
(6.4)
87

For precision oscillators, the second term on the right hand side is quite small and useful to
define the fractional frequency.
yt = vt − v 0
v 0
= 1 d
2 v 0
dt
= dx
dt
(6.5)
where:
xt = t
2 v 0
(6.6)
xt is an expression of the phase in units of time and is thus representative of the total
cumulative time deviation of the reference clock due to instability.
This quantity can be additionally determined by integrating the fractional deviation
respect to time.
yt , with
However, the bandwidth limitations that apply to the measurement of instantaneous frequency
are additionally applicable to instantaneous phase measurement. Therefore, the average
fractional frequency is defined as:
yt =
xt − xt
(6.7)
xt and xt are the respective time deviations measured. is known as the
sampling or averaging time. This is illustrated in the Figure 6.1.
Figure 6.1: Determining the fractional frequency
88

In Figure 6.2, v(t) represents a stable reference oscillator and w(t) represents an oscillator with
instabilities. If viewed over a timescale smaller than t 1 , w(t) is unstable relative to v(t).
However, above this time the oscillators are in phase and as a result, the instability of w(t) would
have gone undetected if time measurements greater than t 1 were made. Furthermore, over the
period t 1 to t 2 , although no instabilities are apparent, when analysed over a smaller time
period, instabilities are noted.
Figure 6.2: Dependence of stability measurement on time window
6.3.1 Measurement of Stability
A variety of methods and their variations exist in order to measure the stability of frequency
sources. The methods most prominent in the literature will be reviewed in this section.
Frequency stability can be measured by taking a reasonably large number of successive readings
of the frequency of the device to be evaluated. Each reading, measured in hertz is obtained by
averaging the output frequency for some specified time period. Furthermore, variations in the
readings of measured frequency might be expected to average out if observed for long enough.
This however, is not always the case [19].
The results are then expressed, often, as a relative or fractional value, given by the following
formula.
=[
F f actual − f ]
nameplate
(6.8)
f nameplate
The nameplate frequency refers to the expected oscillator operating frequency. If an oscillator
operated exactly at its nameplate frequency, it would be considered a perfect frequency source.
However, in practice there is some frequency error and thus a difference f between the
actual and nameplate frequencies. The value of the fractional frequency will become negative if
the measured value is below that of the nominal frequency.
89

The process of dividing the f by the nameplate frequency, normalises the error and allows
frequency stability to be directly manipulated as it is independent of the actual operating
frequency of the oscillator being discussed. For example this makes it possible to compare the
stability of a 10MHz oscillator to a 10kHz oscillator without any additional information.
As a further example, a 1MHz oscillator that is higher in its frequency by 1Hz will have a
fractional frequency equal to 10 -6 or 1 Part Per Million (PPM). The same 1Hz error for a 10MHz
oscillator is a smaller part of the nominal frequency and gives us a smaller relative frequency
equal to 10 -7 or 0.1PPM. Thus, this division of f by the nominal (nameplate) frequency,
removes the concern of weather the nameplate frequency is 1 MHz or 10MHz and is extended to
any other frequency of interest.
In addition to this, this notation allows the measurement of the oscillator output directly or the
use of a divided version of the oscillator. For example, an oscillator with a 1MHz output can
have its output divided to 1Hz without changing the result of its stability. This attribute provides
the ability to deal with sources of any frequency by using dividers to make the measurement
problems more manageable, as it is much easier to deal with lower frequencies, and the final
results are the same.
In describing the second measure of frequency stability, it is useful to mention the relationship
between the time and frequency domains of a signal.
This is understood by use of the continuous Fourier transform pair, defined as:
and
∞
X f = ∫ xt exp[− j 2 f t]dt
−∞
x t = 1 ∫ ∞
X f exp[ j 2 f t]df
2
−∞
(6.9)
(6.10)
The frequency domain representation of the signal xt , the total cumulative time deviation, is
also known as its power spectrum and provides an indication of the distribution of the signal's
power with frequency. The power density spectrum, S x
f , can then be obtained by
normalising the power spectrum, such that the total area under the curve is equated to unity [30].
Furthermore, the power spectrum contains frequency components for both phase and amplitude
fluctuations and when dealing with precision oscillators, the mean squared value of the
amplitude fluctuations are small enough to be neglected. The spectral density, can be then further
scaled to represent the phase power density spectrum, S
f . Finally, S y
f represents
the spectral density of the instantaneous fractional frequency fluctuations yt .
These spectral densities are related in the following manner:
S y
f = f 2
v 0
2
S
f
(6.11)
S x
f =
1
2 v 0
2
S f
S
f = 2 f 2 S
f
90

The third measure of frequency stability is to use the sample variance, y 2 , of the
fractional frequency fluctuations. This variance determines the extent of the variation of the
oscillator's average frequency between two adjacent measurement intervals. If the time or
frequency fluctuations between a pair of oscillators is measured, a process whereby N values of
the fractional frequency y i are measured over a time and measurements are repeated after
an interval of time T . There is a dead time between each frequency measurement, if the
measurement repetition interval is greater than the averaging time and is given by T − , this
is illustrated in Figure 6.3.
The N sample variance can be then computed,
〈 2 1
y
N ,T , 〉 =
〈 ∑ y
N − 1 n
− 1 ∑
n = 1 N k = 1
N
N
2
y k
〉
(6.12)
Figure 6.3: measurement process [30]
The angle brackets in Equation 6.12 denote the infinite time average. Frequently, however, this
result does not converge as N ∞ . This is as a result of noise processes in the oscillator that
diverge at low Fourier frequencies. Thus the precision of the estimation of the variance, does not
simply improve as the sample size is increased. For this reason, the Allan variance is preferred as
a measure of stability and is shown to converge rapidly as the sample size is increased [30]. A
low Allan variance is a characteristic of a clock with good stability of the measured period.
The Allan variance is described as [30]:
y 2 = 〈 1 2 yt − y t 2 〉
(6.13)
91

This equation can be stated equivalently in terms of phase data as [30]:
where:
2 y
=
〈 1 xt 2 − 2xt xt
2
2
2〉
(6.14)
xt is the measured phase difference. This can still furthermore, be manipulated in order to
operate on a discrete time data set as [30]:
y 2 ≃
1
2 N − 2 2
N − 2
∑ x i 2
− 2x i 1
x i
2
i = 1
(6.15)
6.4 System Overview
This section will describe the experimental test setup. The test signal is generated by a Hewlett
Packard 8656B signal generator, with the output frequency set to 100MHz. This signal is split
and sent to the USRP with two TVRx daughterboard modules plugged in and running
simultaneously. This is representative of the setup that will be expected of the PCL receiver with
independent TVRx modules for the reference channel and the channel monitoring target returns.
The independent signals are then digitised by the USRP, synchronisation of the ADCs is ensured
as they are clocked by the same master clock running the USRP. The digitised signals are then
written to file on a PC for offline processing of the data. The functional block diagram of the
system is shown in Figure 6.4 and a photograph of the test equipment is shown in Figure 6.5.
Figure 6.4: Block diagram of experimental setup
92

Figure 6.5: Experiment setup for stability tests
GNURadio was configured such that the signals from tuner 1 and tuner 2 are written to separate
files on the PC, allowing for greater flexibility in post processing. The experiment was
conducted with three variations of the decimation rate, that being 256, 128, and 64. This
translates to data rates varying between 250ksps and 1Msps.
Furthermore, the system was setup in order to determine the oscillator stability over a 30 minute
period. This was achieved by taking successive readings of the frequency output by each tuner
over the 30 minute period. Each reading, was obtained by averaging the output frequency for a 1
second time gate, every minute over the 30 minute test time.
The resolution of the measured frequency is given by:
f = 1 T
(6.16)
Where T is the time gate over which the frequency is averaged.
Furthermore, the phase variation between the signals at the measured intervals was determined
and used in order to verify and confirm the frequency variation that had been determined. A set
of fractional frequencies describing the behaviour of the stability of the oscillator were then
obtained. These results are used in order to generate the Allan deviation of the oscillators.
93

6.5 Results
The following section will present the results of the tests conducted, using the experimental
setup outlined in the above section.
In Figure 6.6, the fractional frequency is observed over the 30 minute time frame. As expected
the crystal ageing effect is gradually reduced over the period, showing a variation from 2.3 to 0.8
parts per 10 million. The large separation initially can be attributed to the innate crystal
oscillator behaviour described as the warm up effect, which is caused by the rise in temperature
of the oscillator from the time the oscillator and instrument is turned on until the time a stable
operating temperature is reached. The temperature rise is brought about directly by power
dissipation in the oscillator or indirectly by the generation of heat in the circuitry of the
instrument. Additionally, the Allan variance of these fractional frequency deviations is shown in
Figure 6.7.
Figure 6.6: Fractional frequency variation of the daughterboards
The Allan variance determines the extent of the variation of the oscillator's average frequency
between two adjacent measurement intervals. Thus, although we observe an overall decline in
the fractional frequency to 0.8 parts per 10 million, the magnitude of the variation between
successive measurements is less severe. The oscillation of the variation indicates movement of
the oscillator frequencies towards and away from each other. The largest measure of the
instantaneous drift is 4Hz. This oscillation can be attributed to physical attributes of the crystal,
losses in the oscillator circuitry, vibration or temperature effects.
94

6.6 Summary
Figure 6.7: Allan Variance of the fractional frequencies
In this chapter it is shown that a critical limiting factor in the performance of radar applications
is the ability of the radar's frequency reference to maintain timing accuracy.
The local oscillator in the TVRx daughterboard of the USRP is a 4MHz quartz crystal oscillator.
It is for this reason that the properties of crystal oscillators and the associated performance
parameters of interest were investigated. These parameters include accuracy, reproducibility and
stability. It was found that crystal oscillators provide good performance at a reasonable price and
dominate the field of frequency sources [19].
Furthermore, the techniques associated with the specification of stability were presented,
including measurement of the frequency deviation and the Allan variance. These tests concluded
that the crystal ageing effect is reduced over the measurement period of thirty minutes, showing
a fractional frequency variation from 2.3 to 0.8 parts per 10 million and a frequency drift no
greater than 4Hz. This falls within the maximum allowance of 5Hz, calculated in section 6.1.
95

Chapter 7
Conclusions and Recommendations
This dissertation provides a comprehensive discussion around bistatic radar with specific
reference to PCL. Various uses of these radar systems, such as military and civilian applications
have been presented, as well as the advantages and disadvantages of the system have been
discussed in detail, and make this type of system desirable for our applications. However, at the
time of writing this dissertation various aspects of the overall project were still under
investigation and as such explicit system requirements were not available, thus, an exhibition
highlighting existing literature and work as well as an examination of the various performance
metrics is made.
In particular the ambiguity function is examined and the performance of commercial FM radio
broadcasts as the radar waveform is determined and summarised by Table 5.2, reproduced here.
Signal
Table 5.2: Summary of waveform ambiguity function
Range Resolution
(km)
Effective
Bandwidth
(kHz)
Peak range
Sidelobe level
(dB)
Peak doppler
Sidelobe level
(dB)
RAD5 3.38 44.3 - 30 - 27
Radio 2000 7.94 18.9 - 18 - 34
Classic FM 7.98 18.8 - 30 - 33
Umhlobe 3.33 45.0 - 31 - 25
RGHP 2.36 63.6 - 30 - 37
RSGR 9.32 16.1 - 18 - 36
SAFM 7.54 19.9 - 20 - 40
Average 5.98 32.37 -25.29 -33.14
The analysis of the FM broadcast signal ambiguity plots reveal the attractive performance of the
signals as they in general tend toward the idealised thumbtack. However, this is greatly
dependant upon the instantaneous modulating content, revealing highly degraded performance in
the case of a signal with low spectral content such as speech with many pauses and breaks.
Furthermore, the SDR paradigm and technology is examined, with discussion around the design
considerations. The USRP, the TVRx daughterboard and GNURadio are examined further as a
potential receiver and development environment, in this light.
96

GNURadio, is found to be a powerful toolkit that implements numerous complex signal
processing elements and additionally interfaces with the USRP that provides flexible and
powerful computing capabilities. Thus combining to provide a powerful and widely versatile
toolset for the development of a variety of radio devices.
The system meets the low cost ambitions costing just over US$500.00 for the USRP
motherboard and a single daughterboard. The receiver's frontend provides a static bandwidth of
6MHz and a tunable range between 50MHz and 800MHz. The noise characterisation of the
receiver reveals a NF of 10dB, a sensitivity of -105dB and a dynamic range of 62dB.
Finally, the investigation into the stability of the daughterboard frontend oscillators due to ageing
effects is shown to be steady through examination of the fractional frequency variation as well as
the Allan variance, showing a fractional frequency variation from 2.3 to 0.8 parts per 10 million
and a frequency drift no greater than 4Hz. This falls within the maximum allowance of 5Hz,
calculated in section 6.1.
In order to further increase the performance of the system, investigation into the use of multiple
waveforms is recommended. The bandwidth of the TVRx lends itself to this. Furthermore, the
effect of multiple USRP receivers and the effect of location and thus multiple simultaneous
locations can be investigated providing a richer information source for the relevant signal
processing.
Special attention should be paid to timing and synchronisation effects and the examination of the
“cross ambiguity function” modelling the bistatic geometry. In addition to this, the versatility of
GNURadio can be further explored, using the inbuilt signal processing blocks to improve the end
to end performance of the PCL system. i.e. The support and performance of adaptive filtering
and various pre-correlation signal conditioning techniques may be explored.
97

Bibliography
[1] Asad A. Abidi, Direct-Conversion Radio Transceiversfor Digital Communications, IEEE
journal of solid-state circuits, Vol 30, no 12, December 1995
[2] Blyakman A.B, Ryndyk A.G, Sidorov S.B, Forward Scattering Radar Moving Object
Coordinate System, Radar, Sonar and Navigation, IEE Proceedings, Volume 152, Issue 3, 3 June
2005 Page(s): 107 - 115
[3] Chang K, RF and Microwave Wireless Systems, Wiley Inter-Science, 2000
[4] Del Mistro M, A Study of Bistatic Radar and the Development of an Independant Bistatic
Receiver, University of Cape Town, 1992
[5] Eric Blossom, GNURadio: Tools for Exploring the Radio Frequency Spectrum,
http://www.linuxjournal.com/article/7319, 2008
[6] Ettus Reasearch LLC, TX and RX Daughterboards for the USRP Software Radio System,
[7] Fette B.A., RF and Wireless Technologies,
[8] Griffiths H D, Garnett A. J., Bistatic radar using satellite-borne illuminators of opportunity,
Radar 92. International Conference, Volume , Issue , 12-13 Oct 1992 Page(s):276 - 279
[9] Griffiths H.D., Bistatic and Multistatic Radar, University College London
[10] Griffiths H.D., Baker C.J., Passive Coherent Location Radar Systems. Part 1: Performance
Prediction, Radar, Sonar and Navigation, IEE Proceedings, Volume 152, Issue 3, 3 June 2005
Page(s): 153 - 159
[11] Griffiths H.D., Baker C.J, Measurement and Analysis of Ambiguity Functions of Passive
Radar Transmissions, Radar Conference, 2005 IEEE International, Volume , Issue , 9-12 May
2005 Page(s): 321 - 325
[12] Griffiths H.D., Long B.A., Television bases Bistatic Radar, IEE. Proceedings, Vol. 133,
Part F, No.7, December 1986 Page(s) 649-657
[13] Guner A., Temple M.A., Claypoole R.L. , Direct path filtering of DAB waveform from PCL
receiver target channel, Electronics Letters, Volume 39, Issue 1, 9 Jan. 2003 Page(s): 118 - 119
[14] Hanle E, Ing Dr, Gunputh H.R., Survey of Bistatic and Multistatic Radar, IEE Proceedings,
133(PtF No7):587– 595,Dec 1986.
[15] Howland P.E. Television Based Bistatic Radar, University of Birmingham,1997
[16] Howland P.E., Maksimiuk D., Reitsma G., FM based Bistatic Radar, Radar, Sonar and
Navigation, IEE Proceedings, Volume 152, Issue 3, 3 June 2005 Page(s): 107 - 115
[17] Jackson M.C, The Geometry of bistatic radar systems, Communications, Radar and Signal
Processing, IEE Proceedings, Volume 133, Issue 7, December 1986 Page(s): 604 - 612
[18] Jiabing Z., Liang T., Yi H., Study on Moving Target Detection to Passive Radar based on
FM broadcast transmitter, Journal of Systems Engineering and Electronics, Volume 18, Issue 3,
2007, Pages 462-468
[19] Kamas G., Lombardi M.A., Time and Frequecy Users Manual, NIST Special publication
559, U.S. Government Printing Office, Washington DC,1990
[20] Mitola J., Software Radio Architecture Evolution: Foundations, Technology Tradeoffs, and
Architecture Implications, IEICE Trans. Commun.E83-B1165–73
[21] Mr Creese, Sentech Transmitter specification,
[22] Nathanson F.E., Radar Design Principles, McGraw-Hill, 1969
[23] Pozar D.M., Microwave and RF Design of Wireless Systems, Wiley, 2001
[24] Quegan S, Kingsley S, Understanding Radar Systems, McGraw-Hill, 1992
[25] Reed J. H., Software Radio: A Modern Approach to Radio Engineering, Prentice Hall, 2002
[26] Sahr J.D., Lind F.D., The Manastash Ridge Radar: a passive bistatic radar for upper
atmospheric radio science, Radio Science, Volume 32, 1997 Page(s) 2345 - 2358
98

[27] Sandenbergh J.S, A GPS synchronised, Quart Crystal Frequency Standard, University of
Cape Town, 2005
[28] schoenenberger, Principles of Independent Receivers for use with Cooperative Radar
Transmitters, The Radio and Electronic Engineer, vol. 52, No. 2, February 1982 Page(s) 93-101
[29] Skolnik M, Introduction to Radar Systems, McGraw-Hill, 1962
[30] Stein S.R., Frequency and Time, Their measurement and characterisation, Precision
Frequency Control, Volume 2, E.A Gerber and A. Ballato, Eds., Academic Press, New
York,1985 Page(s) 191 - 232
[31] The Software Defined Radio Forum, http://www.sdrforum.org, 2008,
[32] Watermeyer K, Design of a hardware platform for narrow-band Software Defined Radio
applications, University of Cape Town, 2007
[33] Wikipedia The Free Encyclopedia, Passive Radar,
http://en.wikipedia.org/wiki/Passive_radar, 2008
[34] Willis N.J., Bistatic Radar, Artech House,1991
[35] Willis N.J., Bistatic radars and their third resurgence: passive coherent location, IEEE
Radar Conference, Long Beach, USA, 2002
[36] Woodward P.M., Probability and Information Theory, with applications to Radar,
Norwood, MA: Artech House, 1980
[37] Microtune, 4937 D15 RF Tuner Module, 3x889 (3x7702), Advance Data Sheet, 2001
[38] Analog Devices, AD9860/AD9862 Datasheet,
[39] GNURadio website, http://www.gnu.org/software/gnuradio, 2008
[40] Chang C.W.W, System Level Investigations of Television Based Bistatic Radar, University
of Cape Town, 2005
99

Appendix A
Ambiguity Analysis Results
Figure A.1: Ambiguity plot of Radio 2000 transmission
100

Figure A.1 a: Zero doppler cut through ambiguity plot
Figure A.1 b: Zero delay cut through ambiguity plot
101

Table A.1: Bandwidth variation of Radio 2000 waveform
Time (s)
Bandwidth (kHz)
0.1 20
0.2 6
0.3 25
0.4 10
0.5 18.5
0.6 31
0.7 14
0.8 5
0.9 19
1.0 15
1.1 33
1.2 0.5
1.3 0.5
1.4 18
1.5 35
1.6 23
1.7 50
1.8 5
1.9 8
2.0 16
Average Bandwidth (kHz) 18.97
102

Figure A.2: Ambiguity plot of Classic FM transmission
Figure A.2 a: Zero doppler cut through ambiguity plot
103

Figure A.2 b: Zero delay cut through ambiguity plot
Table A.2: Bandwidth variation of Classic FM waveform
Time (s)
Bandwidth (kHz)
0.1 9
0.2 7
0.3 20
0.4 25
0.5 25
0.6 15
0.7 24
0.8 22
0.9 9
1.0 32.5
1.1 45
1.2 13
1.3 24
1.4 48
104

1.5 41
1.6 13
1.7 4.5
1.8 0.6
1.9 5
2.0 4.5
Average Bandwidth (kHz) 18.81
Figure A.3: Ambiguity plot of UmhloboFM transmission
105

Figure A.3 a: Zero doppler cut through ambiguity plot
Figure A.3 b: Zero delay cut through ambiguity plot
106

Table A.3: Bandwidth variation of Umhlobo FM waveform
Time (s)
Bandwidth (kHz)
0.1 41
0.2 70
0.3 75
0.4 35
0.5 53
0.6 52
0.7 36
0.8 40
0.9 45
1.0 39
1.1 42.5
1.2 65
1.3 28
1.4 51
1.5 35
1.6 44
1.7 23
1.8 67
1.9 68
2.0 55
Average Bandwidth (kHz) 45
107

Figure A.4: Ambiguity plot for RGHP transmission
Figure A.4 a: Zero doppler cut through ambiguity plot
108

Figure A.4 b: Zero delay cut through ambiguity plot
Table A.4: Bandwidth variation of RGHP waveform
Time (s)
Bandwidth (kHz)
0.1 39
0.2 115
0.3 73
0.4 52
0.5 38
0.6 57
0.7 46
0.8 60
0.9 53
1.0 55
1.1 107
1.2 48
1.3 49
1.4 42
1.5 59
109

1.6 30
1.7 63
1.8 62
1.9 50
2.0 118
Average Bandwidth (kHz) 63.6
Figure A.5: Ambiguity plot of RSGR transmission
110

Figure A.5 a: Zero doppler cut through ambiguity plot
Figure A.5 b: Zero delay cut through ambiguity plot
111

Table A.5: Bandwidth variation of RSGR waveform
Time (s)
Bandwidth (kHz)
0.1 3.5
0.2 0.3
0.3 4
0.4 30
0.5 9
0.6 35
0.7 40
0.8 21
0.9 18
1.0 1.5
1.1 55
1.2 20
1.3 50
1.4 1.2
1.5 4.5
1.6 7.5
1.7 5.5
1.8 14
1.9 5.5
2.0 13
Average Bandwidth (kHz) 16.1
112

Figure A.6: Ambiguity plot of SAFM transmission
113

Figure A.6 a: Zero doppler cut through ambiguity plot
Figure A.6 b: Zero delay cut through ambiguity plot
114

Table A.6: Bandwidth variation of SAFM waveform
Time (s)
Bandwidth (kHz)
0.1 40
0.2 40
0.3 47.5
0.4 30
0.5 47
0.6 13
0.7 17
0.8 38
0.9 2.5
1.0 1.4
1.1 2.5
1.2 6
1.3 18
1.4 40
1.5 6
1.6 7.5
1.7 1.4
1.8 3.5
1.9 1.6
2.0 27.5
Average Bandwidth (kHz) 19.9
115

Appendix B
Data Sheets and Sentech Table
116

STATION NAME
STATION
CODE
LAT
DEG
LAT
MIN
LAT
SEC
LONG
DEG
LONG
MIN
LONG
SEC
Sentech Station Information
(First Page Only)
MAP
NUMBER
SITE
HEIGHT
MAST
HEIGHT
SERVICE
CODE
CAPE TOWN C1 -34 -3 -15 18 23 15 3418AB 902 154 RAD5 89 1 10 FM 10 4 6
CAPE TOWN C1 -34 -3 -15 18 23 15 3418AB 902 154 LOBO 92.1 1 10 FM 10 4 6
CAPE TOWN C1 -34 -3 -15 18 23 15 3418AB 902 154 RGHP 95.3 1 10 FM 10 4 6
CAPE TOWN C1 -34 -3 -15 18 23 15 3418AB 902 154 2000 98.6 1 10 FM 10 4 6
CAPE TOWN C1 -34 -3 -15 18 23 15 3418AB 902 154 RSGR 102.1 1 10 FM 10 4 6
CAPE TOWN C1 -34 -3 -15 18 23 15 3418AB 902 154 SAFM 105.7 1 10 FM 10 4 6
CAPE TOWN C1 -34 -3 -15 18 23 15 3418AB 902 154 SBC1 183.25 1 15.85 TVV 12 4 3
CAPE TOWN C1 -34 -3 -15 18 23 15 3418AB 902 154 SBC2 207.25 1 15.85 TVV 12 4 3
CAPE TOWN C1 -34 -3 -15 18 23 15 3418AB 902 154 MNET 231.25 1 15.85 TVV 12 4 3
CAPE TOWN C1 -34 -3 -15 18 23 15 3418AB 902 154 CSN 735.25 0.1 0.25 TVU 4 4 1
CAPE TOWN C1 -34 -3 -15 18 23 15 3418AB 902 154 ETV 767.25 1 6.76 TVU 8.3 4 4
CAPE TOWN C1 -34 -3 -15 18 23 15 3418AB 902 154 SBC3 799.25 1 6.76 TVU 8.3 4 4
TABLE MOUNTAIN C1.1 -33 -57 -25 18 24 13 3318CD 1067 14 METR 88.6 0.02 0.02 FM 0 1 1
TABLE MOUNTAIN C1.1 -33 -57 -25 18 24 13 3318CD 1067 14 RAD5 89.9 0.02 0.02 FM 0 1 1
TABLE MOUNTAIN C1.1 -33 -57 -25 18 24 13 3318CD 1067 14 LOBO 92.5 0.02 0.02 FM 0 1 1
TABLE MOUNTAIN C1.1 -33 -57 -25 18 24 13 3318CD 1067 14 RGHP 95.8 0.02 0.02 FM 0 1 1
TABLE MOUNTAIN C1.1 -33 -57 -25 18 24 13 3318CD 1067 14 2000 99.1 0.02 0.02 FM 0 1 1
TABLE MOUNTAIN C1.1 -33 -57 -25 18 24 13 3318CD 1067 14 RSGR 102.6 0.02 0.02 FM 0 1 1
TABLE MOUNTAIN C1.1 -33 -57 -25 18 24 13 3318CD 1067 14 SAFM 106.2 0.02 0.02 FM 0 1 1
TABLE MOUNTAIN C1.1 -33 -57 -25 18 24 13 3318CD 1067 14 SBC2 495.25 0.1 0.5 TVU 7 2 1
TABLE MOUNTAIN C1.1 -33 -57 -25 18 24 13 3318CD 1067 14 SBC1 527.25 0.1 0.5 TVU 7 2 1
TABLE MOUNTAIN C1.1 -33 -57 -25 18 24 13 3318CD 1067 14 MNET 591.25 0.1 0.5 TVU 7 2 1
TABLE MOUNTAIN C1.1 -33 -57 -25 18 24 13 3318CD 1067 14 SBC3 751.25 0.1 0.59 TVU 7.7 2 1
TABLE MOUNTAIN C1.1 -33 -57 -25 18 24 13 3318CD 1067 14 CSN 783.25 0.05 0.29 TVU 7.7 2 1
TABLE MOUNTAIN C1.1 -33 -57 -25 18 24 13 3318CD 1067 14 ETV 815.25 0.1 0.59 TVU 7.7 2 1
AURORA C1.10 -33 -49 -39 18 38 29 3318DC 176 23 SBC2 487.25 0 0 TVU 4.2 3 1
AURORA C1.10 -33 -49 -39 18 38 29 3318DC 176 23 SBC3 551.25 0 0 TVU 4.2 3 1
AURORA C1.10 -33 -49 -39 18 38 29 3318DC 176 23 MNET 583.25 0 0 TVU 4.2 3 1
AURORA C1.10 -33 -49 -39 18 38 29 3318DC 176 23 SBC1 727.25 0 0 TVU 4.2 3 1
AURORA C1.10 -33 -49 -39 18 38 29 3318DC 176 23 ETV 759.25 0 0 TVU 4.2 3 1
GRABOUW C1.11 -34 -6 -5 18 58 3 3418BB 1181 30 RHLD 93.6 0.1 0.1 FM 0 1 1
GRABOUW C1.11 -34 -6 -5 18 58 3 3418BB 1181 30 RKFM 94.9 0.01 0.01 FM 0 1 1
GRABOUW C1.11 -34 -6 -5 18 58 3 3418BB 1181 30 RSGR 101.7 0.01 0.01 FM 0 1 1
GRABOUW C1.11 -34 -6 -5 18 58 3 3418BB 1181 30 SAFM 105.3 0.01 0.01 FM 0 1 1
GRABOUW C1.11 -34 -6 -5 18 58 3 3418BB 1181 30 HART 107.8 0 0.01 FM 4 1 1
GRABOUW C1.11 -34 -6 -5 18 58 3 3418BB 1181 30 SBC2 615.25 0.05 0.5 TVU 10 3 4
TX
FREQ
TX OUT
PWR
ERP
SPEC
FREQ
BAND
ANT
GAIN
ANT
FACE
ANT
TIER

M I C R O T U N E
4937 DI5 RF TUNER MODULE
3X8899 (3X7702)
ADVANCE DATA SHEET
CABLE MODEM APPLICATIONS
1 APPLICATIONS
The 4937 DI5 Tuner Module is specifically designed for subscriber-side cable
modem applications.
2 FEATURES
Preliminary Preliminary and Confidential
3 INTRODUCTION
• DOCSIS compatible
• VHF, Hyperband, and UHF
• Band selection and tuning controlled by I 2 C bus
• Downstream frequency range from 50 MHz to 860 MHz
• Upstream frequency range from 5 MHz to 42 MHz
• Single 5V power supply
The receiver uses a single-conversion approach to 43.75 MHz with the reception
frequency range divided into VHF low, VHF high, and UHF. A second conversion to
5.75 MHz is available for QAM demodulators requiring a lower center frequency
(3x7702); alternately, the output frequency is 43.75 MHz (3x8899).
Figure 1
4937 DI5 RF Tuner Modules
Band selection and tuning is done via the I²C-bus, while a separate three-wire bus
and transmit enable control the upstream amplifier.
3X 8899 Revision: 02
September 01
Page 1 of 17
3x8899_02.doc

4937 DI5 TUNER MODULE CABLE MODEM APPLICATIONS
3X8899 (3X7702)
ADVANCE DATA SHEET
The common cable input/output is realized by an F-connector (75Ω) per [IPS-sp-
406].
Two automatic gain control (AGC) inputs are available to level the signal into an
external demodulator. The tuner’s intermediate frequency (IF) output is designed to
drive a low-pass image reject filter prior to the QAM demodulator IC.
A DC/DC converter is built in, so that only a single supply voltage of 5V is required.
4 MECHANICAL SPECIFICATIONS
This section contains mechanical specifications for the 4937 DI5 RF Tuner Module.
4.1 MECHANICAL DRAWING
Preliminary
Figure 2
Mechanical Drawing
3X 8899 Revision: 02
September 01
Page 2 of 17
3x8899_02.doc

4937 DI5 TUNER MODULE CABLE MODEM APPLICATIONS
3X8899 (3X7702)
ADVANCE DATA SHEET
4.2 MECHANICAL CHARACTERISTI CS
Table 1
Mechanical Characteristics
CHARACTERISTIC
DIMENSIONS
Dimensions According to the drawing in Figure 2
Weight
Plug holding strength
Tuner connection
Screw fixing of F-connector ∗
Approximately 56g
Plug according to SCTE spec. IPS-sp-407
The tuner provides four pins at bottom cover for horizontal
mounting and grounding
Absolute maximum torque strength: 3.39 Nm / only once
Absolute maximum cantilever strength: 3.39 Nm
Absolute maximum axial strength: 8.99N
∗
If the tuner is not mounted on the chassis, the frame may be bent during the test.
Regardless of mounting, the F-connector will not be pulled out of the frame.
5 FUNCTIONAL SPECIFICATIONS
Preliminary
5.1 ABSOLUTE MAXIMUM RATING S
Stresses greater than those listed in Table 2 may cause permanent damage to the
device. These are stress ratings only; functional operation of the device under
conditions other than those listed in Table 3 is not recommended or implied.
Exposure to any of the absolute-maximum rating conditions for extended periods of
time may affect reliability.
Table 2
Absolute Maximum Specifications
PARAMETER MIN MAX UNIT
Supply voltage 6 V
AGC voltage 6 V
Storage temperature -30 +70 °C
3X 8899 Revision: 02
September 01
Page 3 of 17
3x8899_02.doc

4937 DI5 TUNER MODULE CABLE MODEM APPLICATIONS
3X8899 (3X7702)
ADVANCE DATA SHEET
5.2 OPERATING CHARACTERISTICS
The operating characteristics listed in Table 3 reflect the conditions necessary for
optimal performance and operating reliability.
Table 3
Operating Characteristics
PARAMETER MIN TYP MAX UNIT
CONDITIONS
OR LOCATION
Frequency range
VHF Low 50 162 MHz
VHF High 156 469 MHz
UHF 463 860 MHz
Frequency range, referenced to center
frequency of 6 MHz bandwidth
VHF Low 53 159 MHz
VHF High 159 466 MHz
UHF 466 857 MHz
Tuning resolution
Preliminary
Standard tuning increment (see
Table 8)
62.5 kHz
Recommended takeover frequencies,
referred to center frequency
VHF Low / VHF High 158 MHz
UHF 464 MHz
Output Frequency
3x7702 5.75 MHz ± 0.05 MHz
3x8899 43.75 MHz ± 0.05 MHz
Input impedance
VHF/UHF Common 75 Ω Unbalanced
AGC voltage for maximum gain
RF 4 V ± 0.1V
IF 4 V ± 0.1V
Power supply voltage
Voltage V S1 5 ± 0.3 V Pin 3
Voltage condition V S1 150 mA
Voltage V S2 5 ±0.25 V Pin 6
Voltage condition V S2 200 mA
Voltage V S3 5 ±0.25 V Pin 10
Voltage condition V S3 200 mA
Voltage V S4 5 ±0.3 V Pin 15
Voltage condition V S4 100 mA
Permissible ripple voltage (20 Hz to 100
kHz)
20 mVpp
3X 8899 Revision: 02
September 01
Page 4 of 17
3x8899_02.doc

4937 DI5 TUNER MODULE CABLE MODEM APPLICATIONS
3X8899 (3X7702)
ADVANCE DATA SHEET
PARAMETER MIN TYP MAX UNIT
CONDITIONS
OR LOCATION
Temperature
Operating temperature 0 60 °C
6 TUNER DOWNSTREAM DATA
Table 4
Electrical Characteristics
PARAMETER TEST CONDITIONS MIN TYP MAX UNIT
Frequency range 55 860 MHz
Preliminary
Input signal level 40 80 dBµV
Voltage gain
Output level at 1 kΩ
Measured between antenna input and
IF output (pins 17 and 18). The input is
loaded with 75Ω and the IF output is
loaded with a test circuit (see Figure 5).
The output impedance is about 220Ω.
Pins 17 and 18 are not DC decoupled.
60 80 95 dB
1 Vpp
VHF Low 8 10 dB
Noise figure
VHF High 8 10 dB
UHF 8 10 dB
VSWR Antenna input 3
IF Rejection
[Rejection of CW
Signal at highest
VHF Low 50 70 dB
possible IF (46.75
MHz) fed into the tuner
input relative to a CW
at desired channel
center frequency
measured at the IF
mixer output. Both
signals must have the
same level at F-
connector input.]
VHF High
UHF
60
60
80
80
dB
dB
Upstream rejection
Image rejection
RF Tilt
Signal level for 1 dB
gain compression
Phase noise
Isolation between upstream output
(5 MHz to 42 MHz) and IF mixer out
(40.75 MHz to 46.75 MHz)
75 dB
VHF Low 60 70 dB
VHF High 55 65 dB
UHF 55 60 dB
For all AGC settings and over a 6 MHz
bandwidth around center frequency
AGC deactivated with AGC = 4V (pins
7 and 16) for maximum gain
2.5 dB
72 dBµV
VHF Low -71 -55 dBc/Hz
VHF High
Measured at 1 kHz distance from
carrier
-60 -55 dBc/Hz
UHF
-58 -55 dBc/Hz
3X 8899 Revision: 02
September 01
Page 5 of 17
3x8899_02.doc

4937 DI5 TUNER MODULE CABLE MODEM APPLICATIONS
3X8899 (3X7702)
ADVANCE DATA SHEET
PARAMETER TEST CONDITIONS MIN TYP MAX UNIT
VHF Low -95 -80 dBc/Hz
VHF High
Measured at 10 kHz distance from
carrier
-85 -80 dBc/Hz
UHF
-85 -80 dBc/Hz
VHF Low -102 -90 dBc/Hz
VHF High
Measured at 20 kHz distance from
carrier
-92 -85 dBc/Hz
UHF
-90 -85 dBc/Hz
VHF Low -109 -100 dBc/Hz
VHF High
Measured at 100 kHz distance from
carrier
-106 -100 dBc/Hz
UHF
-103 -100 dBc/Hz
Oscillator voltage F-connector terminated with 75Ω

4937 DI5 TUNER MODULE CABLE MODEM APPLICATIONS
3X8899 (3X7702)
ADVANCE DATA SHEET
6.1 INFLUENCE OF AGC
The curves in Figure 3 and Figure 4 are measured at +25°C with an input level of
45 dBµV. The values are typical values and can vary within the guaranteed limits.
60,0
50,0
40,0
30,0
RF gain [dB]
20,0
10,0
VHF-low (100MHz)
VHF-high (350MHz)
UHF (650MHz)
0,0
0,0 0,5 1,0 1,5 2,0 2,5 3,0 3,5 4,0 4,5
-10,0
-20,0
RF-AGC voltage [V]
Preliminary
35,0
30,0
25,0
Figure 3
RF Gain vs. AGC Voltage
20,0
IF gain [dB]
15,0
10,0
all bands
5,0
0,0
0,0 0,5 1,0 1,5 2,0 2,5 3,0 3,5 4,0 4,5
-5,0
IF-AGC voltage [V]
Figure 4
IF Gain vs. AGC Voltage
The noise figure shall not increase by more than the corresponding AGC gain
reduction. The input return loss shall be maintained within the specified limits over
the entire range of AGC voltage.
3X 8899 Revision: 02
September 01
Page 7 of 17
3x8899_02.doc

4937 DI5 TUNER MODULE CABLE MODEM APPLICATIONS
3X8899 (3X7702)
ADVANCE DATA SHEET
7 TUNER UPSTREAM DATA
All data is measured according to the test circuit shown in Figure 6 on page 9. The
input impedance between Pins 1 and 2 for this tuner is 50 ohms.
Table 5
Tuner Upstream Data
PARAMETER TEST CONDITIONS MIN TYP MAX UNIT
Input level Source impedance 75Ω sym 33 35 dBmV
Preliminary
Voltage gain
Gain control word = maximum
gain
25 27 29 dB
Gain steps 0.7 1 1.3 dB
Gain range 59 dB
Group delay variation
Amplitude ripple variation
5 MHz to 42 MHz (3.2 MHz
bandwidth)
60 nsec
5 MHz to 42 MHz 1.28 MHz bandwidth ± 0.2 dB
Absolute accuracy of
transmitted power
TX Transient Spurs
Gain setting = maximum
gain
Gain setting < (maximum
gain –12)
5 MHz to 42 MHz ± 2 dB
16 mVp-p
8 mVp-p
TX Transient duration TXEN rise/fall time < 0.1 µs 2 µsec
Reverse channel harmonic
distortion
V out = +58 dBmV
5 MHz to 42 MHz 2 nd harmonic level, single tone -53 dBc
5 MHz to 42 MHz 3 rd harmonic level, single tone -54 dBc
54 MHz to 60 MHz -40 -35 dBmV
60 MHz to 88 MHz -50 -40 dBmV
88 MHz to 860 MHz -50 -45 dBmV
Noise floor
Input terminated with 75Ω
Transmit mode noise Voltage gain 24 dB 131 150 nV / √Hz
Transmit disable mode
noise
TXEN low, voltage gain 24 dB 810 pV / √Hz
8 TUNER MEASUREMENT TEST CO NDITIONS
All tuner data are held under the following conditions unless otherwise noted:
• Measurement tolerance 10% or 1 dB
• Ambient temperature + 25°C ± 3°C
• Supply voltages + 5V ± 2%
• AGC voltage + 4V ± 2%
3X 8899 Revision: 02
September 01
Page 8 of 17
3x8899_02.doc

4937 DI5 TUNER MODULE CABLE MODEM APPLICATIONS
3X8899 (3X7702)
ADVANCE DATA SHEET
8.1 TEST CIRCUITS
8.1.1 VOLTAGE GAIN, TILT, AND NO ISE FIGURE
Tuner
Pin 17
Pin 18
680 Ω
680 =Ω
4:1
T1
75 Ω=Impedance
Figure 5
Test Circuit for Voltage Gain, Tilt, and Noise Figure
Preliminary
For the voltage gain, tilt, and noise figure test circuit:
• Loss of test-dummy: 22.6 dB
• T1 = RF – Transformer (ohms - ratio = 1:4)
• Type: MCL T4-1 or equivalent
8.1.2 UPSTREAM CHANNEL
Tuner
Pin 1
Pin 2
1:1
T1
75 Ω=Impedance
Figure 6
Test Circuit for Upstream Channel
For the upstream channel test circuit:
• Loss of test-dummy: < 1 dB
• T1 = RF – Transformer (ohms - ratio = 1:1)
• Type: MCL T1-1 or equivalent
3X 8899 Revision: 02
September 01
Page 9 of 17
3x8899_02.doc

4937 DI5 TUNER MODULE CABLE MODEM APPLICATIONS
3X8899 (3X7702)
ADVANCE DATA SHEET
9 CONTROL
9.1 WRITE DATA FORMAT FOR I 2 C BUS
Table 6
Write Data Format
MSB LSB ACK
Address byte 1 1 0 0 0 MA1 MA0 R/W 1 A 2
Divider byte 1 0 N14 N13 N12 N11 N10 N9 N8 A
Divider byte 2 N7 N6 N5 N4 N3 N2 N1 N0 A
Control byte 1 1 CP T2 T1 T0 RSA RSB OS A
Control byte 2 P7 P6 P5 P4 P3 P2 P1 P0 A
1
R/W = 0 is write mode
2
A = Acknowledge
9.2 ADDRESS SELECTION FOR I 2 C B US
Preliminary
Table 7 Address Selection
MA1 MA0 ADDRESS VOLTAGE AT PIN 11
0 0 C0 (0 to 0.1) V S3
0 1 C2 Open circuit or (0.2 to 0.3) V S3
1 0 C4 (0.4 to 0.6) V S3
1 1 C6 (0.9 to 1) V S3
9.3 OSCILLATOR FREQUENCY AND DIVIDER BYTE CALCULATION
Table 8
Oscillator Frequency and Divider Byte Calculation
RSA RSB REFERENCE DIVIDER
MINIMUM TUNING
STEP
F REF
1 1 512 62.5 kHz 7.8125 kHz
X 0 640 50.0 kHz 6.25 kHz
0 1 1024 31.25 kHz 3.90625 kHz
Use the following formula to calculate oscillator frequency and divider byte.
f osc = f ref × 8 × SF
3X 8899 Revision: 02
September 01
Page 10 of 17
3x8899_02.doc

4937 DI5 TUNER MODULE CABLE MODEM APPLICATIONS
3X8899 (3X7702)
ADVANCE DATA SHEET
Where:
f osc = Local oscillator frequency
f ref = Crystal reference frequency / 512 = 4 MHz / 512 = 7.8125 kHz
SF = Programmable scaling factor
Scaling factor is SF = 16384 × n14 + 8192 × n13 + 4096 × n12 + 2048 ×
n11 + 1024 × n10 + 512 × n9 + 256 × n8 + 128 × n7 + 64 × n6 + 32 × n5
+ 16 × n4 + 8 × n3 + 4 × n2 + 2 × n1 + n0
9.4 CONTROL BYTE (I 2 C)
Table 9
Control Byte 1 Settings (Default)
MSB LSB ACK
Control byte 1 1 0 0 0 1 1 1 0 A
Table 10
Control Byte 1 Settings Default Descriptions
CODE DESCRIPTION SETTINGS
Preliminary
CP
Charge pump current
1 = Fastest tuning
0 = Better phase noise for distance < 10
kHz to the carrier
OS Tuning voltage
0 = On
1 = Off
RSA, RSB Reference divider See Table 8 on page 10
T0, T1, T2 Test mode bit See Table 11
Table 11
Test Mode Bit Settings
T2 T1 T0 DEVICE OPERATION
0 0 1 Normal mode
0 1 x Charge pump is off
1 1 0 Charge pump is sinking current
1 1 1 Charge pump is sourcing current
1 0 0 Internal test mode
1 0 1 Internal test mode
Table 12
Control Byte 2 (Band Selection)
BAND ACTIVE PORT P7 P6 P5 P4 P3 P2 P1 P0
UHF P0 0 X 1 1 X X X X
VHF High P2 1 X 0 1 X X X X
VHF Low P1 1 X 1 0 X X X X
Note: X = not used, P3 = used for upstream shutdown (see section 9.6)
3X 8899 Revision: 02
September 01
Page 11 of 17
3x8899_02.doc

4937 DI5 TUNER MODULE CABLE MODEM APPLICATIONS
3X8899 (3X7702)
ADVANCE DATA SHEET
9.5 READ DATA FORMAT (I2C)
Table 13 Read Data Format (I 2 C)
MSB LSB ACK
Address byte 1 1 0 0 0 MA1 MA0 R/W A
Status byte POR FL I2 I1 I0 A2 A1 A0 A
Note: MSB is transmitted first.
Table 14
Read Data Format Descriptions
CODE
DESCRIPTION
R/W 1 = Read mode
POR Power on reset flag (POR = 1 at power on)
FL
In lock flag (FL = 1 when PLL is locked)
I2, I1, I0 Digital levels for I/O ports P0, P1, and P2
A2, A1, A0
Digital output of 5-level ADC for AFC function. Values
for correct tuning: A2 = 0, A1= 1, A0 = 0
Preliminary
9.6 PROGRAMMABLE-GAIN AMPLI FIER CONTROL (THREE-WIRE BUS)
Table 15 Pin Map (Three-Wire Bus)
PIN SYMBOL DESCRIPTION
4 AS1 Active low enable
5 TXEnable Hardware shutdown
8 SCL1 Serial clock
9 SDA1 Serial data
A serial data interface controls the programmable-gain amplifier (PGA). It has an
active-low enable (AS1) to sample the data, with data clocked in MSB (D7) first on
the rising edge of SCL1. Data is stored on the rising edge of AS1. The gain is
determined by a 6-bit word (D5 – D0).
Table 16
Data Register (3-Wire Bus)
BIT MNEMONIC DESCRIPTION
MSB 7 D7 Software shutdown
6 D6 Test bit
5 D5 Gain control, bit 5
4 D4 Gain control, bit 4
3 D3 Gain control, bit 3
2 D2 Gain control, bit 2
1 D1 Gain control, bit 1
3X 8899 Revision: 02
September 01
Page 12 of 17
3x8899_02.doc

4937 DI5 TUNER MODULE CABLE MODEM APPLICATIONS
3X8899 (3X7702)
ADVANCE DATA SHEET
BIT MNEMONIC DESCRIPTION
0 D0 Gain control, bit 0
Setting PLL-Port 3 low shuts down the PGA. Port 3 is controlled over the I 2 C bus
(SDA2; SCL2). Control byte 2 (P3) has to be 1 for shutdown or 0 for normal mode.
Hardware shutdown overrides software shutdown (D7) and stored gain settings will
be lost. In normal active mode, port 3 must be held high. To bias only the differential
output-power-amp between bursts, TXEnable (Pin 5) must be held low. TXEnable
must be held high for transmit mode.
Table 17
State Diagram (3-Wire Bus)
SHDN
PORT 3
TXEN
PIN 5 D7 D6 D5 D4 D3 D2 D1 D0 STATE
1 0 X X X X X X X X Shutdown mode
0 0 0 X X X X X X X Software shutdown mode
0 0 1 X X X X X X X Transmit disable mode
0 1 1 X X X X X X X Transmit mode
Preliminary
0 1 1 X 0 0 0 0 0 0
Maximum gain – 63 dB =
minimum gain
0 1 1 X 0 0 0 0 0 1 Maximum gain – 62 dB
0 1 1 X - - - - - - -
0 1 1 X 1 0 0 0 0 1 Maximum gain – 30 dB
0 1 1 X - - - - - - -
0 1 1 X 1 1 1 1 1 0 Maximum gain – 1 dB
0 1 1 X 1 1 1 1 1 1 Maximum gain
9.7 SERIAL INTERFACE TIMING
A G B C D E F
AS1
SCL1
D7 D6 D5 D4 D3 D2 D1 D0
SDA1
Figure 7
Serial Interface Timing
3X 8899 Revision: 02
September 01
Page 13 of 17
3x8899_02.doc

4937 DI5 TUNER MODULE CABLE MODEM APPLICATIONS
3X8899 (3X7702)
ADVANCE DATA SHEET
Table 18
Timing Characteristics
PARAMETER SYMBOL MIN TYP MAX UNITS
AS1 to SCL1 rise setup time A 10 ns
AS1 to SCL1 rise hold time F 20 ns
SDA1 to SCL1 setup time B 10 ns
SDA1 to SCL1 hold time C 20 ns
SDA1 pulse width high G 50 ns
SDA1 pulse width low G 50 ns
SCL1 pulse width high E 50 ns
SCL1 pulse width low D 50 ns
10 SAFETY AND RELIABILITY
10.1 ELECTROSTATIC DISCHARGE (E SD) PROTECTION
Preliminary
10.2 HIGH VOLTAGE
WARNING:
Observe these precautions:
The 4937 DI5 Tuner Module contains components that can be
damaged by electrostatic discharge.
• Ground yourself before handling the tuner.
• Do not touch the tuner connector pins without ESD protection.
The tuner meets specifications IEC 801.2 level 2.
10.3 HUMIDITY
Table 19
Local Oscillator Drift
PARAMETER DRIFT UNIT PROCEDURE
VHF Low ± 15 kHz
VHF High ± 45 kHz
UHF ± 75 kHz
1. Run 60 hours at 55°C and 20% relative
humidity.
2. Run 1 hour at 23°C and 50% relative
humidity.
3. Take first measurement.
4. Run 65 hours at +40°C and 95% relative
humidity.
5. Take second measurement.
3X 8899 Revision: 02
September 01
Page 14 of 17
3x8899_02.doc

4937 DI5 TUNER MODULE CABLE MODEM APPLICATIONS
3X8899 (3X7702)
ADVANCE DATA SHEET
10.4 VIBRATION TEST
After applying vibration of 1.5 mm amplitude, frequency of 10 - 55 -10 Hz (1 minute)
each X, Y, Z direction for 2 hours (total 6 hours), tuner shall not have any rattling or
loosening and shall comply with the variation to its initial value as listed in Table 20.
Table 20
Vibration Test
PARAMETER MEASUREMENT UNIT
Gain variation < ± 3 dB
Wave variation < ± 30 %
10.5 MICROPHONY
The microphony test is made with a TV set. The resolution is optimal. With maximum
AF output of the TV set, the tuner is free of microphonic effects, provided the unit is
installed in a professional manner.
Preliminary
10.6 LOOSE CONTACT TEST OF TUNE R ALONE
10.7 SOLDER LIMITS
The test pattern is a color bar. The resolution is 3 MHz. To test, there must be no
interruption effects when the edge of the tuner is knocked, provided it is fastened
with a ground contact.
See application note APN001.
10.8 NATIONAL REGULATIONS
The tuner meets the requirements of VDE 9872/7.72 and Amtsblatt DBP 069/1981
(FTZ), EN 55013, EN 55020 (if properly mounted into TV set, VCR, or converter).
3X 8899 Revision: 02
September 01
Page 15 of 17
3x8899_02.doc

4937 DI5 TUNER MODULE CABLE MODEM APPLICATIONS
3X8899 (3X7702)
ADVANCE DATA SHEET
11 ORDERING INFORMATION
The 4937 DI5 Tuner Modules may be ordered in the packaging units and quantities
shown in Table 21 and Table 22. For packaging options and quantities other than
those shown, contact one of the offices listed on the last page of this document.
Table 21
Packaging Units
4937 TUNER MODELS
PACKAGING UNITS 3X8899 3X7702
Number of Tuner Modules Per Box 72 72
Number of Boxes Per Master Box 40 40
Table 22
Order Quantities
TOTAL NUMBER OF TUNERS PER
MASTER BOX
NUMBER OF MASTER
BOXES 3X8899 3X7702
Preliminary
0.5 1,440 1,440
1.0 2,880 2,880
1.5 4,320 4,320
2.0 5,760 5,760
2.5 7,200 7,200
3.0 8,640 8,640
3.5 10,080 10,080
4.0 11,520 11,520
4.5 12,960 12,960
5.0 14,400 14,400
12 REVISION HISTORY
NAME
DESCRIPTION
ECN
NO. DATE REV
Hennig 24.11.00 M1
Hennig 011/01 20.02.01 01
Hennig Change 3x7702 (3x8899) to 3x8899 (3x7702) 050/01 10.07.01 02
3X 8899 Revision: 02
September 01
Page 16 of 17
3x8899_02.doc

4937 DI5 TUNER MODULE CABLE MODEM APPLICATIONS
3X8899 (3X7702)
ADVANCE DATA SHEET
NOTICES
Preliminary
Preliminary
and Confidential
NOTICE - The information in this document is believed to be accurate and reliable. Microtune assumes no
responsibility for any consequences arising from the use of this information, nor from any infringement of patents
or the rights of third parties which may result from its use. No license is granted by implication or otherwise under
any patent or other rights of Microtune. The information in this publication replaces and supersedes all
information previously supplied, and is subject to change without notice. The customer is responsible for
assuring that proper design and operating safeguards are observed to minimize inherent and procedural
hazards. Microtune assumes no responsibility for applications assistance or customer product design.
NOTICE - The devices described in this document are not authorized for use in medical, life-support equipment,
or any other application involving a potential risk of severe property or environmental damage, personal injury, or
death without prior express written approval of Microtune. Any such use is understood to be entirely at the user’s
risk.
TRADEMARKS - Microtune, MicroTuner, and the Microtune logo are trademarks of Microtune, Inc. All other
trademarks belong to their respective companies.
PATENTS – Microtune’s products are protected by one or more of the following U.S. patents: 5,625,325;
5,648,744; 5,717,730; 5,737,035; 5,739,730; 5,805,988; 5,847,612; 6,100,761; 6,104,242; 6,144,402; 6,163,684;
6,169,569; 6,177,964; 6,218,899 and additional patents pending or filed.
COPYRIGHT - Entire contents Copyright © 2001 Microtune, Inc.
World Headquarters
Microtune, Inc.
2201 Tenth Street
Plano, TX 75074
USA
Telephone: 972-673-1600
Fax: 972-673-1602
Email: sales@microtune.com
Website: www.microtune.com
European Headquarters
Microtune GmbH and Co. KG
Marie Curie Strasse 1
85055 Ingolstadt / Germany
Telephone: +49-841-9378-011
Fax: +49-841-9378-010
Sales Telephone: +49-841-9378-020
Sales Fax: +49-841-9378-024
Pan-Asian Headquarters
Microtune, Inc. - Hong Kong
Silvercord Tower 1, Room 503
30 Canton Road
Kowloon, Hong Kong
Telephone: +852-2378-8128
Fax: +852-2302-0756
For a detailed list of current sales representatives, visit our Web site at www.microtune.com.
3X 8899 Revision: 02
August 01
Page 17 of 17
3x8899_02.doc